Merge pull request #51 from router-for-me/gemini-web

feat(gemini-web): Add support for real Nano Banana model

.dockerignore

@@ -0,0 +1,32 @@
+# Git and GitHub folders
+.git/*
+.github/*
+
+# Docker and CI/CD related files
+docker-compose.yml
+.dockerignore
+.gitignore
+.goreleaser.yml
+Dockerfile
+
+# Documentation and license
+docs/*
+README.md
+README_CN.md
+MANAGEMENT_API.md
+MANAGEMENT_API_CN.md
+LICENSE
+
+# Example configuration
+config.example.yaml
+
+# Runtime data folders (should be mounted as volumes)
+auths/*
+logs/*
+conv/*
+config.yaml
+
+# Development/editor
+bin/*
+.claude/*
+.vscode/*

.github/ISSUE_TEMPLATE/bug_report.md

@@ -0,0 +1,37 @@
+---
+name: Bug report
+about: Create a report to help us improve
+title: ''
+labels: ''
+assignees: ''
+
+---
+
+**Describe the bug**
+A clear and concise description of what the bug is.
+
+**CLI Type**
+What type of CLI account do you use?  (gemini-cli, gemini, codex, claude code or openai-compatibility)
+
+**Model Name**
+What model are you using? (example: gemini-2.5-pro, claude-sonnet-4-20250514, gpt-5, etc.)
+
+**LLM Client**
+What LLM Client are you using? (example: roo-code, cline, claude code, etc.)
+
+**Request Information**
+The best way is to paste the cURL command of the HTTP request here.
+Alternatively, you can set `request-log: true` in the `config.yaml` file and then upload the detailed log file.
+
+**Expected behavior**
+A clear and concise description of what you expected to happen.
+
+**Screenshots**
+If applicable, add screenshots to help explain your problem.
+
+**OS Type**
+ - OS: [e.g. macOS]
+ - Version [e.g. 15.6.0]
+
+**Additional context**
+Add any other context about the problem here.

.github/workflows/docker-image.yml

@@ -24,8 +24,11 @@ jobs:
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
-      - name: Generate App Version
-        run: echo APP_VERSION=`git describe --tags --always` >> $GITHUB_ENV
+      - name: Generate Build Metadata
+        run: |
+          echo VERSION=`git describe --tags --always --dirty` >> $GITHUB_ENV
+          echo COMMIT=`git rev-parse --short HEAD` >> $GITHUB_ENV
+          echo BUILD_DATE=`date -u +%Y-%m-%dT%H:%M:%SZ` >> $GITHUB_ENV
       - name: Build and push
         uses: docker/build-push-action@v6
         with:
@@ -35,8 +38,9 @@ jobs:
             linux/arm64
           push: true
           build-args: |
-            APP_NAME=${{ env.APP_NAME }}
-            APP_VERSION=${{ env.APP_VERSION }}
+            VERSION=${{ env.VERSION }}
+            COMMIT=${{ env.COMMIT }}
+            BUILD_DATE=${{ env.BUILD_DATE }}
           tags: |
             ${{ env.DOCKERHUB_REPO }}:latest
-            ${{ env.DOCKERHUB_REPO }}:${{ env.APP_VERSION }}
+            ${{ env.DOCKERHUB_REPO }}:${{ env.VERSION }}

.github/workflows/release.yaml

@@ -13,18 +13,26 @@ jobs:
   goreleaser:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           fetch-depth: 0
       - run: git fetch --force --tags
-      - uses: actions/setup-go@v3
+      - uses: actions/setup-go@v4
         with:
           go-version: '>=1.24.0'
           cache: true
-      - uses: goreleaser/goreleaser-action@v3
+      - name: Generate Build Metadata
+        run: |
+          echo VERSION=`git describe --tags --always --dirty` >> $GITHUB_ENV
+          echo COMMIT=`git rev-parse --short HEAD` >> $GITHUB_ENV
+          echo BUILD_DATE=`date -u +%Y-%m-%dT%H:%M:%SZ` >> $GITHUB_ENV
+      - uses: goreleaser/goreleaser-action@v4
         with:
           distribution: goreleaser
           version: latest
           args: release --clean
         env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          VERSION: ${{ env.VERSION }}
+          COMMIT: ${{ env.COMMIT }}
+          BUILD_DATE: ${{ env.BUILD_DATE }}

.gitignore

@@ -1,3 +1,12 @@
 config.yaml
-docs/
-logs/
+bin/*
+docs/*
+logs/*
+conv/*
+auths/*
+!auths/.gitkeep
+.vscode/*
+.claude/*
+AGENTS.md
+CLAUDE.md
+*.exe

.goreleaser.yml

@@ -9,6 +9,8 @@ builds:
       - arm64
     main: ./cmd/server/
     binary: cli-proxy-api
+    ldflags:
+      - -s -w -X 'main.Version={{.Version}}' -X 'main.Commit={{.ShortCommit}}' -X 'main.BuildDate={{.Date}}'
 archives:
   - id: "cli-proxy-api"
     format: tar.gz
@@ -19,4 +21,17 @@ archives:
       - LICENSE
       - README.md
       - README_CN.md
-      - config.example.yaml
+      - config.example.yaml
+
+checksum:
+  name_template: 'checksums.txt'
+
+snapshot:
+  name_template: "{{ incpatch .Version }}-next"
+
+changelog:
+  sort: asc
+  filters:
+    exclude:
+      - '^docs:'
+      - '^test:'

Dockerfile

@@ -8,10 +8,16 @@ RUN go mod download
 
 COPY . .
 
-RUN CGO_ENABLED=0 GOOS=linux go build -o ./CLIProxyAPI ./cmd/server/
+ARG VERSION=dev
+ARG COMMIT=none
+ARG BUILD_DATE=unknown
+
+RUN CGO_ENABLED=0 GOOS=linux go build -ldflags="-s -w -X 'main.Version=${VERSION}' -X 'main.Commit=${COMMIT}' -X 'main.BuildDate=${BUILD_DATE}'" -o ./CLIProxyAPI ./cmd/server/
 
 FROM alpine:3.22.0
 
+RUN apk add --no-cache tzdata
+
 RUN mkdir /CLIProxyAPI
 
 COPY --from=builder ./app/CLIProxyAPI /CLIProxyAPI/CLIProxyAPI
@@ -20,4 +26,8 @@ WORKDIR /CLIProxyAPI
 
 EXPOSE 8317
 
+ENV TZ=Asia/Shanghai
+
+RUN cp /usr/share/zoneinfo/${TZ} /etc/localtime && echo "${TZ}" > /etc/timezone
+
 CMD ["./CLIProxyAPI"]

MANAGEMENT_API.md

@@ -0,0 +1,579 @@
+# Management API
+
+Base path: `http://localhost:8317/v0/management`
+
+This API manages the CLI Proxy API’s runtime configuration and authentication files. All changes are persisted to the YAML config file and hot‑reloaded by the service.
+
+Note: The following options cannot be modified via API and must be set in the config file (restart if needed):
+- `allow-remote-management`
+- `remote-management-key` (if plaintext is detected at startup, it is automatically bcrypt‑hashed and written back to the config)
+
+## Authentication
+
+- All requests (including localhost) must provide a valid management key.
+- Remote access requires enabling remote management in the config: `allow-remote-management: true`.
+- Provide the management key (in plaintext) via either:
+  - `Authorization: Bearer <plaintext-key>`
+  - `X-Management-Key: <plaintext-key>`
+
+If a plaintext key is detected in the config at startup, it will be bcrypt‑hashed and written back to the config file automatically.
+
+## Request/Response Conventions
+
+- Content-Type: `application/json` (unless otherwise noted).
+- Boolean/int/string updates: request body is `{ "value": <type> }`.
+- Array PUT: either a raw array (e.g. `["a","b"]`) or `{ "items": [ ... ] }`.
+- Array PATCH: supports `{ "old": "k1", "new": "k2" }` or `{ "index": 0, "value": "k2" }`.
+- Object-array PATCH: supports matching by index or by key field (specified per endpoint).
+
+## Endpoints
+
+### Config
+- GET `/config` — Get the full config
+    - Request:
+      ```bash
+      curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/config
+      ```
+    - Response:
+      ```json
+      {"debug":true,"proxy-url":"","api-keys":["1...5","JS...W"],"quota-exceeded":{"switch-project":true,"switch-preview-model":true},"generative-language-api-key":["AI...01", "AI...02", "AI...03"],"request-log":true,"request-retry":3,"claude-api-key":[{"api-key":"cr...56","base-url":"https://example.com/api"},{"api-key":"cr...e3","base-url":"http://example.com:3000/api"},{"api-key":"sk-...q2","base-url":"https://example.com"}],"codex-api-key":[{"api-key":"sk...01","base-url":"https://example/v1"}],"openai-compatibility":[{"name":"openrouter","base-url":"https://openrouter.ai/api/v1","api-keys":["sk...01"],"models":[{"name":"moonshotai/kimi-k2:free","alias":"kimi-k2"}]},{"name":"iflow","base-url":"https://apis.iflow.cn/v1","api-keys":["sk...7e"],"models":[{"name":"deepseek-v3.1","alias":"deepseek-v3.1"},{"name":"glm-4.5","alias":"glm-4.5"},{"name":"kimi-k2","alias":"kimi-k2"}]}],"allow-localhost-unauthenticated":true}
+      ```
+
+### Debug
+- GET `/debug` — Get the current debug state
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/debug
+    ```
+  - Response:
+    ```json
+    { "debug": false }
+    ```
+- PUT/PATCH `/debug` — Set debug (boolean)
+  - Request:
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":true}' \
+      http://localhost:8317/v0/management/debug
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+
+### Proxy Server URL
+- GET `/proxy-url` — Get the proxy URL string
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/proxy-url
+    ```
+  - Response:
+    ```json
+    { "proxy-url": "socks5://user:pass@127.0.0.1:1080/" }
+    ```
+- PUT/PATCH `/proxy-url` — Set the proxy URL string
+  - Request (PUT):
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":"socks5://user:pass@127.0.0.1:1080/"}' \
+      http://localhost:8317/v0/management/proxy-url
+    ```
+  - Request (PATCH):
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":"http://127.0.0.1:8080"}' \
+      http://localhost:8317/v0/management/proxy-url
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+- DELETE `/proxy-url` — Clear the proxy URL
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE http://localhost:8317/v0/management/proxy-url
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+
+### Quota Exceeded Behavior
+- GET `/quota-exceeded/switch-project`
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/quota-exceeded/switch-project
+    ```
+  - Response:
+    ```json
+    { "switch-project": true }
+    ```
+- PUT/PATCH `/quota-exceeded/switch-project` — Boolean
+  - Request:
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":false}' \
+      http://localhost:8317/v0/management/quota-exceeded/switch-project
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+- GET `/quota-exceeded/switch-preview-model`
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/quota-exceeded/switch-preview-model
+    ```
+  - Response:
+    ```json
+    { "switch-preview-model": true }
+    ```
+- PUT/PATCH `/quota-exceeded/switch-preview-model` — Boolean
+  - Request:
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":true}' \
+      http://localhost:8317/v0/management/quota-exceeded/switch-preview-model
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+
+### API Keys (proxy service auth)
+- GET `/api-keys` — Return the full list
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/api-keys
+    ```
+  - Response:
+    ```json
+    { "api-keys": ["k1","k2","k3"] }
+    ```
+- PUT `/api-keys` — Replace the full list
+  - Request:
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '["k1","k2","k3"]' \
+      http://localhost:8317/v0/management/api-keys
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+- PATCH `/api-keys` — Modify one item (`old/new` or `index/value`)
+  - Request (by old/new):
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"old":"k2","new":"k2b"}' \
+      http://localhost:8317/v0/management/api-keys
+    ```
+  - Request (by index/value):
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"index":0,"value":"k1b"}' \
+      http://localhost:8317/v0/management/api-keys
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+- DELETE `/api-keys` — Delete one (`?value=` or `?index=`)
+  - Request (by value):
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/api-keys?value=k1'
+    ```
+  - Request (by index):
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/api-keys?index=0'
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+
+### Gemini API Key (Generative Language)
+- GET `/generative-language-api-key`
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/generative-language-api-key
+    ```
+  - Response:
+    ```json
+    { "generative-language-api-key": ["AIzaSy...01","AIzaSy...02"] }
+    ```
+- PUT `/generative-language-api-key`
+  - Request:
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '["AIzaSy-1","AIzaSy-2"]' \
+      http://localhost:8317/v0/management/generative-language-api-key
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+- PATCH `/generative-language-api-key`
+  - Request:
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"old":"AIzaSy-1","new":"AIzaSy-1b"}' \
+      http://localhost:8317/v0/management/generative-language-api-key
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+- DELETE `/generative-language-api-key`
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/generative-language-api-key?value=AIzaSy-2'
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+
+### Codex API KEY (object array)
+- GET `/codex-api-key` — List all
+    - Request:
+      ```bash
+      curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/codex-api-key
+      ```
+    - Response:
+      ```json
+      { "codex-api-key": [ { "api-key": "sk-a", "base-url": "" } ] }
+      ```
+- PUT `/codex-api-key` — Replace the list
+    - Request:
+      ```bash
+      curl -X PUT -H 'Content-Type: application/json' \
+      -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+        -d '[{"api-key":"sk-a"},{"api-key":"sk-b","base-url":"https://c.example.com"}]' \
+        http://localhost:8317/v0/management/codex-api-key
+      ```
+    - Response:
+      ```json
+      { "status": "ok" }
+      ```
+- PATCH `/codex-api-key` — Modify one (by `index` or `match`)
+    - Request (by index):
+      ```bash
+      curl -X PATCH -H 'Content-Type: application/json' \
+      -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+        -d '{"index":1,"value":{"api-key":"sk-b2","base-url":"https://c.example.com"}}' \
+        http://localhost:8317/v0/management/codex-api-key
+      ```
+    - Request (by match):
+      ```bash
+      curl -X PATCH -H 'Content-Type: application/json' \
+      -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+        -d '{"match":"sk-a","value":{"api-key":"sk-a","base-url":""}}' \
+        http://localhost:8317/v0/management/codex-api-key
+      ```
+    - Response:
+      ```json
+      { "status": "ok" }
+      ```
+- DELETE `/codex-api-key` — Delete one (`?api-key=` or `?index=`)
+    - Request (by api-key):
+      ```bash
+      curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/codex-api-key?api-key=sk-b2'
+      ```
+    - Request (by index):
+      ```bash
+      curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/codex-api-key?index=0'
+      ```
+    - Response:
+      ```json
+      { "status": "ok" }
+      ```
+
+### Request Retry Count
+- GET `/request-retry` — Get integer
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/request-retry
+    ```
+  - Response:
+    ```json
+    { "request-retry": 3 }
+    ```
+- PUT/PATCH `/request-retry` — Set integer
+  - Request:
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":5}' \
+      http://localhost:8317/v0/management/request-retry
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+
+### Allow Localhost Unauthenticated
+- GET `/allow-localhost-unauthenticated` — Get boolean
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/allow-localhost-unauthenticated
+    ```
+  - Response:
+    ```json
+    { "allow-localhost-unauthenticated": false }
+    ```
+- PUT/PATCH `/allow-localhost-unauthenticated` — Set boolean
+  - Request:
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":true}' \
+      http://localhost:8317/v0/management/allow-localhost-unauthenticated
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+
+### Claude API KEY (object array)
+- GET `/claude-api-key` — List all
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/claude-api-key
+    ```
+  - Response:
+    ```json
+    { "claude-api-key": [ { "api-key": "sk-a", "base-url": "" } ] }
+    ```
+- PUT `/claude-api-key` — Replace the list
+  - Request:
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '[{"api-key":"sk-a"},{"api-key":"sk-b","base-url":"https://c.example.com"}]' \
+      http://localhost:8317/v0/management/claude-api-key
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+- PATCH `/claude-api-key` — Modify one (by `index` or `match`)
+  - Request (by index):
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"index":1,"value":{"api-key":"sk-b2","base-url":"https://c.example.com"}}' \
+      http://localhost:8317/v0/management/claude-api-key
+    ```
+  - Request (by match):
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"match":"sk-a","value":{"api-key":"sk-a","base-url":""}}' \
+      http://localhost:8317/v0/management/claude-api-key
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+- DELETE `/claude-api-key` — Delete one (`?api-key=` or `?index=`)
+  - Request (by api-key):
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/claude-api-key?api-key=sk-b2'
+    ```
+  - Request (by index):
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/claude-api-key?index=0'
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+
+### OpenAI Compatibility Providers (object array)
+- GET `/openai-compatibility` — List all
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/openai-compatibility
+    ```
+  - Response:
+    ```json
+    { "openai-compatibility": [ { "name": "openrouter", "base-url": "https://openrouter.ai/api/v1", "api-keys": [], "models": [] } ] }
+    ```
+- PUT `/openai-compatibility` — Replace the list
+  - Request:
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '[{"name":"openrouter","base-url":"https://openrouter.ai/api/v1","api-keys":["sk"],"models":[{"name":"m","alias":"a"}]}]' \
+      http://localhost:8317/v0/management/openai-compatibility
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+- PATCH `/openai-compatibility` — Modify one (by `index` or `name`)
+  - Request (by name):
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"name":"openrouter","value":{"name":"openrouter","base-url":"https://openrouter.ai/api/v1","api-keys":[],"models":[]}}' \
+      http://localhost:8317/v0/management/openai-compatibility
+    ```
+  - Request (by index):
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"index":0,"value":{"name":"openrouter","base-url":"https://openrouter.ai/api/v1","api-keys":[],"models":[]}}' \
+      http://localhost:8317/v0/management/openai-compatibility
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+- DELETE `/openai-compatibility` — Delete (`?name=` or `?index=`)
+  - Request (by name):
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/openai-compatibility?name=openrouter'
+    ```
+  - Request (by index):
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/openai-compatibility?index=0'
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+
+### Auth File Management
+
+Manage JSON token files under `auth-dir`: list, download, upload, delete.
+
+- GET `/auth-files` — List
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/auth-files
+    ```
+  - Response:
+    ```json
+    { "files": [ { "name": "acc1.json", "size": 1234, "modtime": "2025-08-30T12:34:56Z", "type": "google" } ] }
+    ```
+
+- GET `/auth-files/download?name=<file.json>` — Download a single file
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -OJ 'http://localhost:8317/v0/management/auth-files/download?name=acc1.json'
+    ```
+
+- POST `/auth-files` — Upload
+  - Request (multipart):
+    ```bash
+    curl -X POST -F 'file=@/path/to/acc1.json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      http://localhost:8317/v0/management/auth-files
+    ```
+  - Request (raw JSON):
+    ```bash
+    curl -X POST -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d @/path/to/acc1.json \
+      'http://localhost:8317/v0/management/auth-files?name=acc1.json'
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+
+- DELETE `/auth-files?name=<file.json>` — Delete a single file
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/auth-files?name=acc1.json'
+    ```
+  - Response:
+    ```json
+    { "status": "ok" }
+    ```
+
+- DELETE `/auth-files?all=true` — Delete all `.json` files under `auth-dir`
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/auth-files?all=true'
+    ```
+  - Response:
+    ```json
+    { "status": "ok", "deleted": 3 }
+    ```
+
+### Login/OAuth URLs
+
+These endpoints initiate provider login flows and return a URL to open in a browser. Tokens are saved under `auths/` once the flow completes.
+
+- GET `/anthropic-auth-url` — Start Anthropic (Claude) login
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      http://localhost:8317/v0/management/anthropic-auth-url
+    ```
+  - Response:
+    ```json
+    { "status": "ok", "url": "https://..." }
+    ```
+
+- GET `/codex-auth-url` — Start Codex login
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      http://localhost:8317/v0/management/codex-auth-url
+    ```
+  - Response:
+    ```json
+    { "status": "ok", "url": "https://..." }
+    ```
+
+- GET `/gemini-cli-auth-url` — Start Google (Gemini CLI) login
+  - Query params:
+    - `project_id` (optional): Google Cloud project ID.
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      'http://localhost:8317/v0/management/gemini-cli-auth-url?project_id=<PROJECT_ID>'
+    ```
+  - Response:
+    ```json
+    { "status": "ok", "url": "https://..." }
+    ```
+
+- GET `/qwen-auth-url` — Start Qwen login (device flow)
+  - Request:
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      http://localhost:8317/v0/management/qwen-auth-url
+    ```
+  - Response:
+    ```json
+    { "status": "ok", "url": "https://..." }
+    ```
+
+## Error Responses
+
+Generic error format:
+- 400 Bad Request: `{ "error": "invalid body" }`
+- 401 Unauthorized: `{ "error": "missing management key" }` or `{ "error": "invalid management key" }`
+- 403 Forbidden: `{ "error": "remote management disabled" }`
+- 404 Not Found: `{ "error": "item not found" }` or `{ "error": "file not found" }`
+- 500 Internal Server Error: `{ "error": "failed to save config: ..." }`
+
+## Notes
+
+- Changes are written back to the YAML config file and hot‑reloaded by the file watcher and clients.
+- `allow-remote-management` and `remote-management-key` cannot be changed via the API; configure them in the config file.

MANAGEMENT_API_CN.md

@@ -0,0 +1,579 @@
+# 管理 API
+
+基础路径：`http://localhost:8317/v0/management`
+
+该 API 用于管理 CLI Proxy API 的运行时配置与认证文件。所有变更会持久化写入 YAML 配置文件，并由服务自动热重载。
+
+注意：以下选项不能通过 API 修改，需在配置文件中设置（如有必要可重启）：
+- `allow-remote-management`
+- `remote-management-key`（若在启动时检测到明文，会自动进行 bcrypt 加密并写回配置）
+
+## 认证
+
+- 所有请求（包括本地访问）都必须提供有效的管理密钥.
+- 远程访问需要在配置文件中开启远程访问： `allow-remote-management: true`
+- 通过以下任意方式提供管理密钥（明文）：
+  - `Authorization: Bearer <plaintext-key>`
+  - `X-Management-Key: <plaintext-key>`
+
+若在启动时检测到配置中的管理密钥为明文，会自动使用 bcrypt 加密并回写到配置文件中。
+
+## 请求/响应约定
+
+- Content-Type：`application/json`（除非另有说明）。
+- 布尔/整数/字符串更新：请求体为 `{ "value": <type> }`。
+- 数组 PUT：既可使用原始数组（如 `["a","b"]`），也可使用 `{ "items": [ ... ] }`。
+- 数组 PATCH：支持 `{ "old": "k1", "new": "k2" }` 或 `{ "index": 0, "value": "k2" }`。
+- 对象数组 PATCH：支持按索引或按关键字段匹配（各端点中单独说明）。
+
+## 端点说明
+
+### Config
+- GET `/config` — 获取完整的配置
+    - 请求:
+      ```bash
+      curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/config
+      ```
+    - 响应:
+      ```json
+      {"debug":true,"proxy-url":"","api-keys":["1...5","JS...W"],"quota-exceeded":{"switch-project":true,"switch-preview-model":true},"generative-language-api-key":["AI...01", "AI...02", "AI...03"],"request-log":true,"request-retry":3,"claude-api-key":[{"api-key":"cr...56","base-url":"https://example.com/api"},{"api-key":"cr...e3","base-url":"http://example.com:3000/api"},{"api-key":"sk-...q2","base-url":"https://example.com"}],"codex-api-key":[{"api-key":"sk...01","base-url":"https://example/v1"}],"openai-compatibility":[{"name":"openrouter","base-url":"https://openrouter.ai/api/v1","api-keys":["sk...01"],"models":[{"name":"moonshotai/kimi-k2:free","alias":"kimi-k2"}]},{"name":"iflow","base-url":"https://apis.iflow.cn/v1","api-keys":["sk...7e"],"models":[{"name":"deepseek-v3.1","alias":"deepseek-v3.1"},{"name":"glm-4.5","alias":"glm-4.5"},{"name":"kimi-k2","alias":"kimi-k2"}]}],"allow-localhost-unauthenticated":true}
+      ```
+
+### Debug
+- GET `/debug` — 获取当前 debug 状态
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/debug
+    ```
+  - 响应：
+    ```json
+    { "debug": false }
+    ```
+- PUT/PATCH `/debug` — 设置 debug（布尔值）
+  - 请求：
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":true}' \
+      http://localhost:8317/v0/management/debug
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+
+### 代理服务器 URL
+- GET `/proxy-url` — 获取代理 URL 字符串
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/proxy-url
+    ```
+  - 响应：
+    ```json
+    { "proxy-url": "socks5://user:pass@127.0.0.1:1080/" }
+    ```
+- PUT/PATCH `/proxy-url` — 设置代理 URL 字符串
+  - 请求（PUT）：
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":"socks5://user:pass@127.0.0.1:1080/"}' \
+      http://localhost:8317/v0/management/proxy-url
+    ```
+  - 请求（PATCH）：
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":"http://127.0.0.1:8080"}' \
+      http://localhost:8317/v0/management/proxy-url
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+- DELETE `/proxy-url` — 清空代理 URL
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE http://localhost:8317/v0/management/proxy-url
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+
+### 超出配额行为
+- GET `/quota-exceeded/switch-project`
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/quota-exceeded/switch-project
+    ```
+  - 响应：
+    ```json
+    { "switch-project": true }
+    ```
+- PUT/PATCH `/quota-exceeded/switch-project` — 布尔值
+  - 请求：
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":false}' \
+      http://localhost:8317/v0/management/quota-exceeded/switch-project
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+- GET `/quota-exceeded/switch-preview-model`
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/quota-exceeded/switch-preview-model
+    ```
+  - 响应：
+    ```json
+    { "switch-preview-model": true }
+    ```
+- PUT/PATCH `/quota-exceeded/switch-preview-model` — 布尔值
+  - 请求：
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":true}' \
+      http://localhost:8317/v0/management/quota-exceeded/switch-preview-model
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+
+### API Keys（代理服务认证）
+- GET `/api-keys` — 返回完整列表
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/api-keys
+    ```
+  - 响应：
+    ```json
+    { "api-keys": ["k1","k2","k3"] }
+    ```
+- PUT `/api-keys` — 完整改写列表
+  - 请求：
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '["k1","k2","k3"]' \
+      http://localhost:8317/v0/management/api-keys
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+- PATCH `/api-keys` — 修改其中一个（`old/new` 或 `index/value`）
+  - 请求（按 old/new）：
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"old":"k2","new":"k2b"}' \
+      http://localhost:8317/v0/management/api-keys
+    ```
+  - 请求（按 index/value）：
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"index":0,"value":"k1b"}' \
+      http://localhost:8317/v0/management/api-keys
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+- DELETE `/api-keys` — 删除其中一个（`?value=` 或 `?index=`）
+  - 请求（按值删除）：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/api-keys?value=k1'
+    ```
+  - 请求（按索引删除）：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/api-keys?index=0'
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+
+### Gemini API Key（生成式语言）
+- GET `/generative-language-api-key`
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/generative-language-api-key
+    ```
+  - 响应：
+    ```json
+    { "generative-language-api-key": ["AIzaSy...01","AIzaSy...02"] }
+    ```
+- PUT `/generative-language-api-key`
+  - 请求：
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '["AIzaSy-1","AIzaSy-2"]' \
+      http://localhost:8317/v0/management/generative-language-api-key
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+- PATCH `/generative-language-api-key`
+  - 请求：
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"old":"AIzaSy-1","new":"AIzaSy-1b"}' \
+      http://localhost:8317/v0/management/generative-language-api-key
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+- DELETE `/generative-language-api-key`
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/generative-language-api-key?value=AIzaSy-2'
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+
+### Codex API KEY（对象数组）
+- GET `/codex-api-key` — 列出全部
+    - 请求：
+      ```bash
+      curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/codex-api-key
+      ```
+    - 响应：
+      ```json
+      { "codex-api-key": [ { "api-key": "sk-a", "base-url": "" } ] }
+      ```
+- PUT `/codex-api-key` — 完整改写列表
+    - 请求：
+      ```bash
+      curl -X PUT -H 'Content-Type: application/json' \
+      -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+        -d '[{"api-key":"sk-a"},{"api-key":"sk-b","base-url":"https://c.example.com"}]' \
+        http://localhost:8317/v0/management/codex-api-key
+      ```
+    - 响应：
+      ```json
+      { "status": "ok" }
+      ```
+- PATCH `/codex-api-key` — 修改其中一个（按 `index` 或 `match`）
+    - 请求（按索引）：
+      ```bash
+      curl -X PATCH -H 'Content-Type: application/json' \
+      -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+        -d '{"index":1,"value":{"api-key":"sk-b2","base-url":"https://c.example.com"}}' \
+        http://localhost:8317/v0/management/codex-api-key
+      ```
+    - 请求（按匹配）：
+      ```bash
+      curl -X PATCH -H 'Content-Type: application/json' \
+      -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+        -d '{"match":"sk-a","value":{"api-key":"sk-a","base-url":""}}' \
+        http://localhost:8317/v0/management/codex-api-key
+      ```
+    - 响应：
+      ```json
+      { "status": "ok" }
+      ```
+- DELETE `/codex-api-key` — 删除其中一个（`?api-key=` 或 `?index=`）
+    - 请求（按 api-key）：
+      ```bash
+      curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/codex-api-key?api-key=sk-b2'
+      ```
+    - 请求（按索引）：
+      ```bash
+      curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/codex-api-key?index=0'
+      ```
+    - 响应：
+      ```json
+      { "status": "ok" }
+      ```
+
+### 请求重试次数
+- GET `/request-retry` — 获取整数
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/request-retry
+    ```
+  - 响应：
+    ```json
+    { "request-retry": 3 }
+    ```
+- PUT/PATCH `/request-retry` — 设置整数
+  - 请求：
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":5}' \
+      http://localhost:8317/v0/management/request-retry
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+
+### 允许本地未认证访问
+- GET `/allow-localhost-unauthenticated` — 获取布尔值
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/allow-localhost-unauthenticated
+    ```
+  - 响应：
+    ```json
+    { "allow-localhost-unauthenticated": false }
+    ```
+- PUT/PATCH `/allow-localhost-unauthenticated` — 设置布尔值
+  - 请求：
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"value":true}' \
+      http://localhost:8317/v0/management/allow-localhost-unauthenticated
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+
+### Claude API KEY（对象数组）
+- GET `/claude-api-key` — 列出全部
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/claude-api-key
+    ```
+  - 响应：
+    ```json
+    { "claude-api-key": [ { "api-key": "sk-a", "base-url": "" } ] }
+    ```
+- PUT `/claude-api-key` — 完整改写列表
+  - 请求：
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '[{"api-key":"sk-a"},{"api-key":"sk-b","base-url":"https://c.example.com"}]' \
+      http://localhost:8317/v0/management/claude-api-key
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+- PATCH `/claude-api-key` — 修改其中一个（按 `index` 或 `match`）
+  - 请求（按索引）：
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"index":1,"value":{"api-key":"sk-b2","base-url":"https://c.example.com"}}' \
+      http://localhost:8317/v0/management/claude-api-key
+    ```
+  - 请求（按匹配）：
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"match":"sk-a","value":{"api-key":"sk-a","base-url":""}}' \
+      http://localhost:8317/v0/management/claude-api-key
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+- DELETE `/claude-api-key` — 删除其中一个（`?api-key=` 或 `?index=`）
+  - 请求（按 api-key）：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/claude-api-key?api-key=sk-b2'
+    ```
+  - 请求（按索引）：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/claude-api-key?index=0'
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+
+### OpenAI 兼容提供商（对象数组）
+- GET `/openai-compatibility` — 列出全部
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/openai-compatibility
+    ```
+  - 响应：
+    ```json
+    { "openai-compatibility": [ { "name": "openrouter", "base-url": "https://openrouter.ai/api/v1", "api-keys": [], "models": [] } ] }
+    ```
+- PUT `/openai-compatibility` — 完整改写列表
+  - 请求：
+    ```bash
+    curl -X PUT -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '[{"name":"openrouter","base-url":"https://openrouter.ai/api/v1","api-keys":["sk"],"models":[{"name":"m","alias":"a"}]}]' \
+      http://localhost:8317/v0/management/openai-compatibility
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+- PATCH `/openai-compatibility` — 修改其中一个（按 `index` 或 `name`）
+  - 请求（按名称）：
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"name":"openrouter","value":{"name":"openrouter","base-url":"https://openrouter.ai/api/v1","api-keys":[],"models":[]}}' \
+      http://localhost:8317/v0/management/openai-compatibility
+    ```
+  - 请求（按索引）：
+    ```bash
+    curl -X PATCH -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d '{"index":0,"value":{"name":"openrouter","base-url":"https://openrouter.ai/api/v1","api-keys":[],"models":[]}}' \
+      http://localhost:8317/v0/management/openai-compatibility
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+- DELETE `/openai-compatibility` — 删除（`?name=` 或 `?index=`）
+  - 请求（按名称）：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/openai-compatibility?name=openrouter'
+    ```
+  - 请求（按索引）：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/openai-compatibility?index=0'
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+
+### 认证文件管理
+
+管理 `auth-dir` 下的 JSON 令牌文件：列出、下载、上传、删除。
+
+- GET `/auth-files` — 列表
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' http://localhost:8317/v0/management/auth-files
+    ```
+  - 响应：
+    ```json
+    { "files": [ { "name": "acc1.json", "size": 1234, "modtime": "2025-08-30T12:34:56Z", "type": "google" } ] }
+    ```
+
+- GET `/auth-files/download?name=<file.json>` — 下载单个文件
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -OJ 'http://localhost:8317/v0/management/auth-files/download?name=acc1.json'
+    ```
+
+- POST `/auth-files` — 上传
+  - 请求（multipart）：
+    ```bash
+    curl -X POST -F 'file=@/path/to/acc1.json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      http://localhost:8317/v0/management/auth-files
+    ```
+  - 请求（原始 JSON）：
+    ```bash
+    curl -X POST -H 'Content-Type: application/json' \
+    -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      -d @/path/to/acc1.json \
+      'http://localhost:8317/v0/management/auth-files?name=acc1.json'
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+
+- DELETE `/auth-files?name=<file.json>` — 删除单个文件
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/auth-files?name=acc1.json'
+    ```
+  - 响应：
+    ```json
+    { "status": "ok" }
+    ```
+
+- DELETE `/auth-files?all=true` — 删除 `auth-dir` 下所有 `.json` 文件
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' -X DELETE 'http://localhost:8317/v0/management/auth-files?all=true'
+    ```
+  - 响应：
+    ```json
+    { "status": "ok", "deleted": 3 }
+    ```
+
+### 登录/授权 URL
+
+以下端点用于发起各提供商的登录流程，并返回需要在浏览器中打开的 URL。流程完成后，令牌会保存到 `auths/` 目录。
+
+- GET `/anthropic-auth-url` — 开始 Anthropic（Claude）登录
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      http://localhost:8317/v0/management/anthropic-auth-url
+    ```
+  - 响应：
+    ```json
+    { "status": "ok", "url": "https://..." }
+    ```
+
+- GET `/codex-auth-url` — 开始 Codex 登录
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      http://localhost:8317/v0/management/codex-auth-url
+    ```
+  - 响应：
+    ```json
+    { "status": "ok", "url": "https://..." }
+    ```
+
+- GET `/gemini-cli-auth-url` — 开始 Google（Gemini CLI）登录
+  - 查询参数：
+    - `project_id`（可选）：Google Cloud 项目 ID。
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      'http://localhost:8317/v0/management/gemini-cli-auth-url?project_id=<PROJECT_ID>'
+    ```
+  - 响应：
+    ```json
+    { "status": "ok", "url": "https://..." }
+    ```
+
+- GET `/qwen-auth-url` — 开始 Qwen 登录（设备授权流程）
+  - 请求：
+    ```bash
+    curl -H 'Authorization: Bearer <MANAGEMENT_KEY>' \
+      http://localhost:8317/v0/management/qwen-auth-url
+    ```
+  - 响应：
+    ```json
+    { "status": "ok", "url": "https://..." }
+    ```
+
+## 错误响应
+
+通用错误格式：
+- 400 Bad Request: `{ "error": "invalid body" }`
+- 401 Unauthorized: `{ "error": "missing management key" }` 或 `{ "error": "invalid management key" }`
+- 403 Forbidden: `{ "error": "remote management disabled" }`
+- 404 Not Found: `{ "error": "item not found" }` 或 `{ "error": "file not found" }`
+- 500 Internal Server Error: `{ "error": "failed to save config: ..." }`
+
+## 说明
+
+- 变更会写回 YAML 配置文件，并由文件监控器热重载配置与客户端。
+- `allow-remote-management` 与 `remote-management-key` 不能通过 API 修改，需在配置文件中设置。

README.md

@@ -2,11 +2,11 @@
 
 English | [中文](README_CN.md)
 
-A proxy server that provides OpenAI/Gemini/Claude compatible API interfaces for CLI.
+A proxy server that provides OpenAI/Gemini/Claude/Codex compatible API interfaces for CLI.
 
 It now also supports OpenAI Codex (GPT models) and Claude Code via OAuth.
 
-So you can use local or multi-account CLI access with OpenAI-compatible clients and SDKs.
+So you can use local or multi-account CLI access with OpenAI(include Responses)/Gemini/Claude-compatible clients and SDKs.
 
 The first Chinese provider has now been added: [Qwen Code](https://github.com/QwenLM/qwen-code).
 
@@ -16,6 +16,7 @@ The first Chinese provider has now been added: [Qwen Code](https://github.com/Qw
 - OpenAI Codex support (GPT models) via OAuth login
 - Claude Code support via OAuth login
 - Qwen Code support via OAuth login
+- Gemini Web support via cookie-based login
 - Streaming and non-streaming responses
 - Function calling/tools support
 - Multimodal input support (text and images)
@@ -25,6 +26,7 @@ The first Chinese provider has now been added: [Qwen Code](https://github.com/Qw
 - Gemini CLI multi-account load balancing
 - Claude Code multi-account load balancing
 - Qwen Code multi-account load balancing
+- OpenAI Codex multi-account load balancing
 - OpenAI-compatible upstream providers via config (e.g., OpenRouter)
 
 ## Installation
@@ -46,9 +48,16 @@ The first Chinese provider has now been added: [Qwen Code](https://github.com/Qw
    ```
 
 2. Build the application:
+   
+   Linux, macOS:
    ```bash
    go build -o cli-proxy-api ./cmd/server
    ```
+   Windows: 
+   ```bash
+   go build -o cli-proxy-api.exe ./cmd/server
+   ```
+
 
 ## Usage
 
@@ -68,6 +77,13 @@ You can authenticate for Gemini, OpenAI, and/or Claude. All can coexist in the s
 
   Options: add `--no-browser` to print the login URL instead of opening a browser. The local OAuth callback uses port `8085`.
 
+- Gemini Web (via Cookies):
+  This method authenticates by simulating a browser, using cookies obtained from the Gemini website.
+  ```bash
+  ./cli-proxy-api --gemini-web-auth
+  ```
+  You will be prompted to enter your `__Secure-1PSID` and `__Secure-1PSIDTS` values. Please retrieve these cookies from your browser's developer tools.
+
 - OpenAI (Codex/GPT via OAuth):
   ```bash
   ./cli-proxy-api --codex-login
@@ -219,7 +235,9 @@ console.log(await claudeResponse.json());
 
 - gemini-2.5-pro
 - gemini-2.5-flash
+- gemini-2.5-flash-lite
 - gpt-5
+- gpt-5-codex
 - claude-opus-4-1-20250805
 - claude-opus-4-20250514
 - claude-sonnet-4-20250514
@@ -239,28 +257,40 @@ The server uses a YAML configuration file (`config.yaml`) located in the project
 
 ### Configuration Options
 
-| Parameter                               | Type     | Default            | Description                                                                                             |
-|-----------------------------------------|----------|--------------------|---------------------------------------------------------------------------------------------------------|
-| `port`                                  | integer  | 8317               | The port number on which the server will listen.                                                        |
-| `auth-dir`                              | string   | "~/.cli-proxy-api" | Directory where authentication tokens are stored. Supports using `~` for the home directory.            |
-| `proxy-url`                             | string   | ""                 | Proxy URL. Supports socks5/http/https protocols. Example: socks5://user:pass@192.168.1.1:1080/           |
-| `request-retry`                         | integer  | 0                  | Number of times to retry a request. Retries will occur if the HTTP response code is 403, 408, 500, 502, 503, or 504. |
-| `quota-exceeded`                        | object   | {}                 | Configuration for handling quota exceeded.                                                              |
-| `quota-exceeded.switch-project`         | boolean  | true               | Whether to automatically switch to another project when a quota is exceeded.                            |
-| `quota-exceeded.switch-preview-model`   | boolean  | true               | Whether to automatically switch to a preview model when a quota is exceeded.                            |
-| `debug`                                 | boolean  | false              | Enable debug mode for verbose logging.                                                                  |
-| `api-keys`                              | string[] | []                 | List of API keys that can be used to authenticate requests.                                             |
-| `generative-language-api-key`           | string[] | []                 | List of Generative Language API keys.                                                                   |
-| `claude-api-key`                        | object   | {}                 | List of Claude API keys.                                                                                |
-| `claude-api-key.api-key`                | string   | ""                 | Claude API key.                                                                                         |
-| `claude-api-key.base-url`               | string   | ""                 | Custom Claude API endpoint, if you use a third-party API endpoint.                                      |
-| `openai-compatibility`                  | object[] | []                 | Upstream OpenAI-compatible providers configuration (name, base-url, api-keys, models).                  |
-| `openai-compatibility.*.name`           | string   | ""                 | The name of the provider. It will be used in the user agent and other places.                           |
-| `openai-compatibility.*.base-url`       | string   | ""                 | The base URL of the provider.                                                                           |
-| `openai-compatibility.*.api-keys`       | string[] | []                 | The API keys for the provider. Add multiple keys if needed. Omit if unauthenticated access is allowed. |
-| `openai-compatibility.*.models`         | object[] | []                 | The actual model name.                                                                                  |
-| `openai-compatibility.*.models.*.name`  | string   | ""                 | The models supported by the provider.                                                                   |
-| `openai-compatibility.*.models.*.alias` | string   | ""                 | The alias used in the API.                                                                              |
+| Parameter                               | Type     | Default            | Description                                                                                                                                                                               |
+|-----------------------------------------|----------|--------------------|-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
+| `port`                                  | integer  | 8317               | The port number on which the server will listen.                                                                                                                                          |
+| `auth-dir`                              | string   | "~/.cli-proxy-api" | Directory where authentication tokens are stored. Supports using `~` for the home directory. If you use Windows, please set the directory like this: `C:/cli-proxy-api/`                  |
+| `proxy-url`                             | string   | ""                 | Proxy URL. Supports socks5/http/https protocols. Example: socks5://user:pass@192.168.1.1:1080/                                                                                            |
+| `request-retry`                         | integer  | 0                  | Number of times to retry a request. Retries will occur if the HTTP response code is 403, 408, 500, 502, 503, or 504.                                                                      |
+| `remote-management.allow-remote`        | boolean  | false              | Whether to allow remote (non-localhost) access to the management API. If false, only localhost can access. A management key is still required for localhost.                              |
+| `remote-management.secret-key`          | string   | ""                 | Management key. If a plaintext value is provided, it will be hashed on startup using bcrypt and persisted back to the config file. If empty, the entire management API is disabled (404). |
+| `quota-exceeded`                        | object   | {}                 | Configuration for handling quota exceeded.                                                                                                                                                |
+| `quota-exceeded.switch-project`         | boolean  | true               | Whether to automatically switch to another project when a quota is exceeded.                                                                                                              |
+| `quota-exceeded.switch-preview-model`   | boolean  | true               | Whether to automatically switch to a preview model when a quota is exceeded.                                                                                                              |
+| `debug`                                 | boolean  | false              | Enable debug mode for verbose logging.                                                                                                                                                    |
+| `api-keys`                              | string[] | []                 | List of API keys that can be used to authenticate requests.                                                                                                                               |
+| `generative-language-api-key`           | string[] | []                 | List of Generative Language API keys.                                                                                                                                                     |
+| `force-gpt-5-codex`                     | bool     | false              | Force the conversion of GPT-5 calls to GPT-5 Codex.                                                                                                                                       |
+| `codex-api-key`                         | object   | {}                 | List of Codex API keys.                                                                                                                                                                   |
+| `codex-api-key.api-key`                 | string   | ""                 | Codex API key.                                                                                                                                                                            |
+| `codex-api-key.base-url`                | string   | ""                 | Custom Codex API endpoint, if you use a third-party API endpoint.                                                                                                                         |
+| `claude-api-key`                        | object   | {}                 | List of Claude API keys.                                                                                                                                                                  |
+| `claude-api-key.api-key`                | string   | ""                 | Claude API key.                                                                                                                                                                           |
+| `claude-api-key.base-url`               | string   | ""                 | Custom Claude API endpoint, if you use a third-party API endpoint.                                                                                                                        |
+| `openai-compatibility`                  | object[] | []                 | Upstream OpenAI-compatible providers configuration (name, base-url, api-keys, models).                                                                                                    |
+| `openai-compatibility.*.name`           | string   | ""                 | The name of the provider. It will be used in the user agent and other places.                                                                                                             |
+| `openai-compatibility.*.base-url`       | string   | ""                 | The base URL of the provider.                                                                                                                                                             |
+| `openai-compatibility.*.api-keys`       | string[] | []                 | The API keys for the provider. Add multiple keys if needed. Omit if unauthenticated access is allowed.                                                                                    |
+| `openai-compatibility.*.models`         | object[] | []                 | The actual model name.                                                                                                                                                                    |
+| `openai-compatibility.*.models.*.name`  | string   | ""                 | The models supported by the provider.                                                                                                                                                     |
+| `openai-compatibility.*.models.*.alias` | string   | ""                 | The alias used in the API.                                                                                                                                                                |
+| `gemini-web`                            | object   | {}                 | Configuration specific to the Gemini Web client.                                                                                                                                          |
+| `gemini-web.context`                    | boolean  | true               | Enables conversation context reuse for continuous dialogue.                                                                                                                               |
+| `gemini-web.code-mode`                  | boolean  | false              | Enables code mode for optimized responses in coding-related tasks.                                                                                                                        |
+| `gemini-web.max-chars-per-request`      | integer  | 1,000,000          | The maximum number of characters to send to Gemini Web in a single request.                                                                                                               |
+| `gemini-web.disable-continuation-hint`  | boolean  | false              | Disables the continuation hint for split prompts.                                                                                                                                         |
+| `gemini-web.token-refresh-seconds`      | integer  | 540                | The interval in seconds for background cookie auto-refresh.                                                                                                                               |
 
 ### Example Configuration File
 
@@ -268,7 +298,18 @@ The server uses a YAML configuration file (`config.yaml`) located in the project
 # Server port
 port: 8317
 
-# Authentication directory (supports ~ for home directory)
+# Management API settings
+remote-management:
+  # Whether to allow remote (non-localhost) management access.
+  # When false, only localhost can access management endpoints (a key is still required).
+  allow-remote: false
+
+  # Management key. If a plaintext value is provided here, it will be hashed on startup.
+  # All management requests (even from localhost) require this key.
+  # Leave empty to disable the Management API entirely (404 for all /v0/management routes).
+  secret-key: ""
+
+# Authentication directory (supports ~ for home directory). If you use Windows, please set the directory like this: `C:/cli-proxy-api/`
 auth-dir: "~/.cli-proxy-api"
 
 # Enable debug logging
@@ -285,6 +326,13 @@ quota-exceeded:
    switch-project: true # Whether to automatically switch to another project when a quota is exceeded
    switch-preview-model: true # Whether to automatically switch to a preview model when a quota is exceeded
 
+# Gemini Web client configuration
+gemini-web:
+  context: true # Enable conversation context reuse
+  code-mode: false # Enable code mode
+  max-chars-per-request: 1000000 # Max characters per request
+  token-refresh-seconds: 540 # Cookie refresh interval in seconds
+
 # API keys for authentication
 api-keys:
   - "your-api-key-1"
@@ -296,6 +344,14 @@ generative-language-api-key:
   - "AIzaSy...02"
   - "AIzaSy...03"
   - "AIzaSy...04"
+
+# Force the conversion of GPT-5 calls to GPT-5 Codex.
+force-gpt-5-codex: true
+
+# Codex API keys
+codex-api-key:
+  - api-key: "sk-atSM..."
+    base-url: "https://www.example.com" # use the custom codex API endpoint
   
 # Claude API keys
 claude-api-key:
@@ -393,12 +449,20 @@ export ANTHROPIC_MODEL=gemini-2.5-pro
 export ANTHROPIC_SMALL_FAST_MODEL=gemini-2.5-flash
 ```
 
-Using OpenAI models:
+Using OpenAI GPT 5 models:
 ```bash
 export ANTHROPIC_BASE_URL=http://127.0.0.1:8317
 export ANTHROPIC_AUTH_TOKEN=sk-dummy
 export ANTHROPIC_MODEL=gpt-5
-export ANTHROPIC_SMALL_FAST_MODEL=gpt-5-nano
+export ANTHROPIC_SMALL_FAST_MODEL=gpt-5-minimal
+```
+
+Using OpenAI GPT 5 Codex models:
+```bash
+export ANTHROPIC_BASE_URL=http://127.0.0.1:8317
+export ANTHROPIC_AUTH_TOKEN=sk-dummy
+export ANTHROPIC_MODEL=gpt-5-codex
+export ANTHROPIC_SMALL_FAST_MODEL=gpt-5-codex-low
 ```
 
 Using Claude models:
@@ -417,6 +481,29 @@ export ANTHROPIC_MODEL=qwen3-coder-plus
 export ANTHROPIC_SMALL_FAST_MODEL=qwen3-coder-flash
 ```
 
+## Codex with multiple account load balancing
+
+Start CLI Proxy API server, and then edit the `~/.codex/config.toml` and `~/.codex/auth.json` files.
+
+config.toml:
+```toml
+model_provider = "cliproxyapi"
+model = "gpt-5-codex" # Or gpt-5, you can also use any of the models that we support.
+model_reasoning_effort = "high"
+
+[model_providers.cliproxyapi]
+name = "cliproxyapi"
+base_url = "http://127.0.0.1:8317/v1"
+wire_api = "responses"
+```
+
+auth.json:
+```json
+{
+  "OPENAI_API_KEY": "sk-dummy"
+}
+```
+
 ## Run with Docker
 
 Run the following command to login (Gemini OAuth on port 8085): 
@@ -425,16 +512,28 @@ Run the following command to login (Gemini OAuth on port 8085):
 docker run --rm -p 8085:8085 -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest /CLIProxyAPI/CLIProxyAPI --login
 ```
 
+Run the following command to login (Gemini Web Cookies):
+
+```bash
+docker run -it --rm -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest /CLIProxyAPI/CLIProxyAPI --gemini-web-auth
+```
+
 Run the following command to login (OpenAI OAuth on port 1455):
 
 ```bash
 docker run --rm -p 1455:1455 -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest /CLIProxyAPI/CLIProxyAPI --codex-login
 ```
 
-Run the following command to login (Claude OAuth on port 54545):
+Run the following command to logi (Claude OAuth on port 54545):
 
 ```bash
-docker run --rm -p 54545:54545 -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest /CLIProxyAPI/CLIProxyAPI --claude-login
+docker run -rm -p 54545:54545 -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest /CLIProxyAPI/CLIProxyAPI --claude-login
+```
+
+Run the following command to login (Qwen OAuth):
+
+```bash
+docker run -it -rm -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest /CLIProxyAPI/CLIProxyAPI --qwen-login
 ```
 
 Run the following command to start the server:
@@ -443,6 +542,77 @@ Run the following command to start the server:
 docker run --rm -p 8317:8317 -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest
 ```
 
+## Run with Docker Compose
+
+1.  Clone the repository and navigate into the directory:
+    ```bash
+    git clone https://github.com/luispater/CLIProxyAPI.git
+    cd CLIProxyAPI
+    ```
+
+2.  Prepare the configuration file:
+    Create a `config.yaml` file by copying the example and customize it to your needs.
+    ```bash
+    cp config.example.yaml config.yaml
+    ```
+    *(Note for Windows users: You can use `copy config.example.yaml config.yaml` in CMD or PowerShell.)*
+
+3.  Start the service:
+    -   **For most users (recommended):**
+        Run the following command to start the service using the pre-built image from Docker Hub. The service will run in the background.
+        ```bash
+        docker compose up -d
+        ```
+    -   **For advanced users:**
+        If you have modified the source code and need to build a new image, use the interactive helper scripts:
+        -   For Windows (PowerShell):
+            ```powershell
+            .\docker-build.ps1
+            ```
+        -   For Linux/macOS:
+            ```bash
+            bash docker-build.sh
+            ```
+        The script will prompt you to choose how to run the application:
+        - **Option 1: Run using Pre-built Image (Recommended)**: Pulls the latest official image from the registry and starts the container. This is the easiest way to get started.
+        - **Option 2: Build from Source and Run (For Developers)**: Builds the image from the local source code, tags it as `cli-proxy-api:local`, and then starts the container. This is useful if you are making changes to the source code.
+
+4. To authenticate with providers, run the login command inside the container:
+    - **Gemini**: 
+    ```bash
+    docker compose exec cli-proxy-api /CLIProxyAPI/CLIProxyAPI -no-browser --login
+    ```
+    - **Gemini Web**:
+    ```bash
+    docker compose exec cli-proxy-api /CLIProxyAPI/CLIProxyAPI --gemini-web-auth
+    ```
+    - **OpenAI (Codex)**:
+    ```bash
+    docker compose exec cli-proxy-api /CLIProxyAPI/CLIProxyAPI -no-browser --codex-login
+    ```
+    - **Claude**: 
+    ```bash
+    docker compose exec cli-proxy-api /CLIProxyAPI/CLIProxyAPI -no-browser --claude-login
+    ```
+    - **Qwen**: 
+    ```bash
+    docker compose exec cli-proxy-api /CLIProxyAPI/CLIProxyAPI -no-browser --qwen-login
+    ```
+
+5.  To view the server logs:
+    ```bash
+    docker compose logs -f
+    ```
+
+6.  To stop the application:
+    ```bash
+    docker compose down
+    ```
+
+## Management API
+
+see [MANAGEMENT_API.md](MANAGEMENT_API.md)
+
 ## Contributing
 
 Contributions are welcome! Please feel free to submit a Pull Request.

README_CN.md

@@ -1,21 +1,42 @@
+# 写给所有中国网友的
+
+对于项目前期的确有很多用户使用上遇到各种各样的奇怪问题，大部分是因为配置或我说明文档不全导致的。
+
+对说明文档我已经尽可能的修补，有些重要的地方我甚至已经写到了打包的配置文件里。
+
+已经写在 README 中的功能，都是**可用**的，经过**验证**的，并且我自己**每天**都在使用的。
+
+可能在某些场景中使用上效果并不是很出色，但那基本上是模型和工具的原因，比如用 Claude Code 的时候，有的模型就无法正确使用工具，比如 Gemini，就在 Claude Code 和 Codex 的下使用的相当扭捏，有时能完成大部分工作，但有时候却只说不做。
+
+目前来说 Claude 和 GPT-5 是目前使用各种第三方CLI工具运用的最好的模型，我自己也是多个账号做均衡负载使用。
+
+实事求是的说，最初的几个版本我根本就没有中文文档，我至今所有文档也都是使用英文更新让后让 Gemini 翻译成中文的。但是无论如何都不会出现中文文档无法理解的问题。因为所有的中英文文档我都是再三校对，并且发现未及时更改的更新的地方都快速更新掉了。
+
+最后，烦请在发 Issue 之前请认真阅读这篇文档。
+
+另外中文需要交流的用户可以加 QQ 群：188637136
+
+或 Telegram 群：https://t.me/CLIProxyAPI
+
 # CLI 代理 API
 
 [English](README.md) | 中文
 
-一个为 CLI 提供 OpenAI/Gemini/Claude 兼容 API 接口的代理服务器。
+一个为 CLI 提供 OpenAI/Gemini/Claude/Codex 兼容 API 接口的代理服务器。
 
 现已支持通过 OAuth 登录接入 OpenAI Codex（GPT 系列）和 Claude Code。
 
-您可以使用本地或多账户的CLI方式，通过任何与OpenAI兼容的客户端和SDK进行访问。
+您可以使用本地或多账户的CLI方式，通过任何与 OpenAI（包括Responses）/Gemini/Claude 兼容的客户端和SDK进行访问。
 
 现已新增首个中国提供商：[Qwen Code](https://github.com/QwenLM/qwen-code)。
 
 ## 功能特性
 
-- 为 CLI 模型提供 OpenAI/Gemini/Claude 兼容的 API 端点
+- 为 CLI 模型提供 OpenAI/Gemini/Claude/Codex 兼容的 API 端点
 - 新增 OpenAI Codex（GPT 系列）支持（OAuth 登录）
 - 新增 Claude Code 支持（OAuth 登录）
 - 新增 Qwen Code 支持（OAuth 登录）
+- 新增 Gemini Web 支持（通过 Cookie 登录）
 - 支持流式与非流式响应
 - 函数调用/工具支持
 - 多模态输入（文本、图片）
@@ -25,6 +46,7 @@
 - 支持 Gemini CLI 多账户轮询
 - 支持 Claude Code 多账户轮询
 - 支持 Qwen Code 多账户轮询
+- 支持 OpenAI Codex 多账户轮询
 - 通过配置接入上游 OpenAI 兼容提供商（例如 OpenRouter）
 
 ## 安装
@@ -68,6 +90,13 @@
 
   选项：加上 `--no-browser` 可打印登录地址而不自动打开浏览器。本地 OAuth 回调端口为 `8085`。
 
+- Gemini Web (通过 Cookie):
+  此方法通过模拟浏览器行为，使用从 Gemini 网站获取的 Cookie 进行身份验证。
+  ```bash
+  ./cli-proxy-api --gemini-web-auth
+  ```
+  程序将提示您输入 `__Secure-1PSID` 和 `__Secure-1PSIDTS` 的值。请从您的浏览器开发者工具中获取这些 Cookie。
+
 - OpenAI（Codex/GPT，OAuth）：
   ```bash
   ./cli-proxy-api --codex-login
@@ -218,7 +247,9 @@ console.log(await claudeResponse.json());
 
 - gemini-2.5-pro
 - gemini-2.5-flash
+- gemini-2.5-flash-lite
 - gpt-5
+- gpt-5-codex
 - claude-opus-4-1-20250805
 - claude-opus-4-20250514
 - claude-sonnet-4-20250514
@@ -233,33 +264,45 @@ console.log(await claudeResponse.json());
 服务器默认使用位于项目根目录的 YAML 配置文件（`config.yaml`）。您可以使用 `--config` 标志指定不同的配置文件路径：
 
 ```bash
-./cli-proxy-api --config /path/to/your/config.yaml
+  ./cli-proxy-api --config /path/to/your/config.yaml
 ```
 
 ### 配置选项
 
-| 参数                                    | 类型       | 默认值                | 描述                                                                                          |
-|---------------------------------------|----------|--------------------|---------------------------------------------------------------------------------------------|
-| `port`                                | integer  | 8317               | 服务器将监听的端口号。                                                                          |
-| `auth-dir`                            | string   | "~/.cli-proxy-api" | 存储身份验证令牌的目录。支持使用 `~` 来表示主目录。                                                  |
-| `proxy-url`                           | string   | ""                 | 代理URL。支持socks5/http/https协议。例如：socks5://user:pass@192.168.1.1:1080/                  |
-| `request-retry`                       | integer  | 0                  | 请求重试次数。如果HTTP响应码为403、408、500、502、503或504，将会触发重试。                                |
-| `quota-exceeded`                      | object   | {}                 | 用于处理配额超限的配置。                                                                        |
-| `quota-exceeded.switch-project`       | boolean  | true               | 当配额超限时，是否自动切换到另一个项目。                                                              |
-| `quota-exceeded.switch-preview-model` | boolean  | true               | 当配额超限时，是否自动切换到预览模型。                                                                |
-| `debug`                               | boolean  | false              | 启用调试模式以获取详细日志。                                                                    |
-| `api-keys`                            | string[] | []                 | 可用于验证请求的API密钥列表。                                                                   |
-| `generative-language-api-key`         | string[] | []                 | 生成式语言API密钥列表。                                                                         |
-| `claude-api-key`                      | object   | {}                 | Claude API密钥列表。                                                                            |
-| `claude-api-key.api-key`              | string   | ""                 | Claude API密钥。                                                                                |
-| `claude-api-key.base-url`             | string   | ""                 | 自定义的Claude API端点，如果您使用第三方的API端点。                                                 |
-| `openai-compatibility`                | object[] | []                 | 上游OpenAI兼容提供商的配置（名称、基础URL、API密钥、模型）。                                        |
-| `openai-compatibility.*.name`           | string   | ""                 | 提供商的名称。它将被用于用户代理（User Agent）和其他地方。                                            |
-| `openai-compatibility.*.base-url`       | string   | ""                 | 提供商的基础URL。                                                                               |
-| `openai-compatibility.*.api-keys`       | string[] | []                 | 提供商的API密钥。如果需要，可以添加多个密钥。如果允许未经身份验证的访问，则可以省略。                         |
-| `openai-compatibility.*.models`         | object[] | []                 | 实际的模型名称。                                                                                |
-| `openai-compatibility.*.models.*.name`  | string   | ""                 | 提供商支持的模型。                                                                              |
-| `openai-compatibility.*.models.*.alias` | string   | ""                 | 在API中使用的别名。                                                                             |
+| 参数                                      | 类型       | 默认值                | 描述                                                                  |
+|-----------------------------------------|----------|--------------------|---------------------------------------------------------------------|
+| `port`                                  | integer  | 8317               | 服务器将监听的端口号。                                                         |
+| `auth-dir`                              | string   | "~/.cli-proxy-api" | 存储身份验证令牌的目录。支持使用 `~` 来表示主目录。如果你使用Windows，建议设置成`C:/cli-proxy-api/`。  |
+| `proxy-url`                             | string   | ""                 | 代理URL。支持socks5/http/https协议。例如：socks5://user:pass@192.168.1.1:1080/ |
+| `request-retry`                         | integer  | 0                  | 请求重试次数。如果HTTP响应码为403、408、500、502、503或504，将会触发重试。                    |
+| `remote-management.allow-remote`        | boolean  | false              | 是否允许远程（非localhost）访问管理接口。为false时仅允许本地访问；本地访问同样需要管理密钥。               |
+| `remote-management.secret-key`          | string   | ""                 | 管理密钥。若配置为明文，启动时会自动进行bcrypt加密并写回配置文件。若为空，管理接口整体不可用（404）。             |
+| `quota-exceeded`                        | object   | {}                 | 用于处理配额超限的配置。                                                        |
+| `quota-exceeded.switch-project`         | boolean  | true               | 当配额超限时，是否自动切换到另一个项目。                                                |
+| `quota-exceeded.switch-preview-model`   | boolean  | true               | 当配额超限时，是否自动切换到预览模型。                                                 |
+| `debug`                                 | boolean  | false              | 启用调试模式以获取详细日志。                                                      |
+| `api-keys`                              | string[] | []                 | 可用于验证请求的API密钥列表。                                                    |
+| `generative-language-api-key`           | string[] | []                 | 生成式语言API密钥列表。                                                       |
+| `force-gpt-5-codex`                     | bool     | false              | 强制将 GPT-5 调用转换成 GPT-5 Codex。                                        |
+| `codex-api-key`                         | object   | {}                 | Codex API密钥列表。                                                      |
+| `codex-api-key.api-key`                 | string   | ""                 | Codex API密钥。                                                        |
+| `codex-api-key.base-url`                | string   | ""                 | 自定义的Codex API端点                                                     |
+| `claude-api-key`                        | object   | {}                 | Claude API密钥列表。                                                     |
+| `claude-api-key.api-key`                | string   | ""                 | Claude API密钥。                                                       |
+| `claude-api-key.base-url`               | string   | ""                 | 自定义的Claude API端点，如果您使用第三方的API端点。                                    |
+| `openai-compatibility`                  | object[] | []                 | 上游OpenAI兼容提供商的配置（名称、基础URL、API密钥、模型）。                                |
+| `openai-compatibility.*.name`           | string   | ""                 | 提供商的名称。它将被用于用户代理（User Agent）和其他地方。                                  |
+| `openai-compatibility.*.base-url`       | string   | ""                 | 提供商的基础URL。                                                          |
+| `openai-compatibility.*.api-keys`       | string[] | []                 | 提供商的API密钥。如果需要，可以添加多个密钥。如果允许未经身份验证的访问，则可以省略。                        |
+| `openai-compatibility.*.models`         | object[] | []                 | 实际的模型名称。                                                            |
+| `openai-compatibility.*.models.*.name`  | string   | ""                 | 提供商支持的模型。                                                           |
+| `openai-compatibility.*.models.*.alias` | string   | ""                 | 在API中使用的别名。                                                         |
+| `gemini-web`                            | object   | {}                 | Gemini Web 客户端的特定配置。                                                 |
+| `gemini-web.context`                    | boolean  | true               | 是否启用会话上下文重用，以实现连续对话。                                        |
+| `gemini-web.code-mode`                  | boolean  | false              | 是否启用代码模式，优化代码相关任务的响应。                                      |
+| `gemini-web.max-chars-per-request`      | integer  | 1,000,000          | 单次请求发送给 Gemini Web 的最大字符数。                                        |
+| `gemini-web.disable-continuation-hint`  | boolean  | false              | 当提示被拆分时，是否禁用连续提示的暗示。                                        |
+| `gemini-web.token-refresh-seconds`      | integer  | 540                | 后台 Cookie 自动刷新的间隔（秒）。                                            |
 
 ### 配置文件示例
 
@@ -267,7 +310,17 @@ console.log(await claudeResponse.json());
 # 服务器端口
 port: 8317
 
-# 身份验证目录（支持 ~ 表示主目录）
+# 管理 API 设置
+remote-management:
+  # 是否允许远程（非localhost）访问管理接口。为false时仅允许本地访问（但本地访问同样需要管理密钥）。
+  allow-remote: false
+
+  # 管理密钥。若配置为明文，启动时会自动进行bcrypt加密并写回配置文件。
+  # 所有管理请求（包括本地）都需要该密钥。
+  # 若为空，/v0/management 整体处于 404（禁用）。
+  secret-key: ""
+
+# 身份验证目录（支持 ~ 表示主目录）。如果你使用Windows，建议设置成`C:/cli-proxy-api/`。
 auth-dir: "~/.cli-proxy-api"
 
 # 启用调试日志
@@ -285,6 +338,13 @@ quota-exceeded:
    switch-project: true # 当配额超限时是否自动切换到另一个项目
    switch-preview-model: true # 当配额超限时是否自动切换到预览模型
 
+# Gemini Web 客户端配置
+gemini-web:
+  context: true # 启用会话上下文重用
+  code-mode: false # 启用代码模式
+  max-chars-per-request: 1000000 # 单次请求最大字符数
+  token-refresh-seconds: 540 # Cookie 刷新间隔（秒）
+
 # 用于本地身份验证的 API 密钥
 api-keys:
   - "your-api-key-1"
@@ -297,11 +357,19 @@ generative-language-api-key:
   - "AIzaSy...03"
   - "AIzaSy...04"
 
-# Claude API keys
-claude-api-key:
-  - api-key: "sk-atSM..." # use the official claude API key, no need to set the base url
+# 强制将 GPT-5 调用转换成 GPT-5 Codex.
+force-gpt-5-codex: true
+
+# Codex API 密钥
+codex-api-key:
   - api-key: "sk-atSM..."
-    base-url: "https://www.example.com" # use the custom claude API endpoint
+    base-url: "https://www.example.com" # 第三方 Codex API 中转服务端点
+
+# Claude API 密钥
+claude-api-key:
+  - api-key: "sk-atSM..." # 如果使用官方 Claude API，无需设置 base-url
+  - api-key: "sk-atSM..."
+    base-url: "https://www.example.com" # 第三方 Claude API 中转服务端点
 
 # OpenAI 兼容提供商
 openai-compatibility:
@@ -388,14 +456,23 @@ export ANTHROPIC_MODEL=gemini-2.5-pro
 export ANTHROPIC_SMALL_FAST_MODEL=gemini-2.5-flash
 ```
 
-使用 OpenAI 模型：
+使用 OpenAI GPT 5 模型：
 ```bash
 export ANTHROPIC_BASE_URL=http://127.0.0.1:8317
 export ANTHROPIC_AUTH_TOKEN=sk-dummy
 export ANTHROPIC_MODEL=gpt-5
-export ANTHROPIC_SMALL_FAST_MODEL=gpt-5-nano
+export ANTHROPIC_SMALL_FAST_MODEL=gpt-5-minimal
 ```
 
+使用 OpenAI GPT 5 Codex 模型:
+```bash
+export ANTHROPIC_BASE_URL=http://127.0.0.1:8317
+export ANTHROPIC_AUTH_TOKEN=sk-dummy
+export ANTHROPIC_MODEL=gpt-5-codex
+export ANTHROPIC_SMALL_FAST_MODEL=gpt-5-codex-low
+```
+
+
 使用 Claude 模型：
 ```bash
 export ANTHROPIC_BASE_URL=http://127.0.0.1:8317
@@ -412,6 +489,28 @@ export ANTHROPIC_MODEL=qwen3-coder-plus
 export ANTHROPIC_SMALL_FAST_MODEL=qwen3-coder-flash
 ```
 
+## Codex 多账户负载均衡
+
+启动 CLI Proxy API 服务器, 修改 `~/.codex/config.toml` 和 `~/.codex/auth.json` 文件。
+
+config.toml:
+```toml
+model_provider = "cliproxyapi"
+model = "gpt-5-codex" # 或者是gpt-5，你也可以使用任何我们支持的模型
+model_reasoning_effort = "high"
+
+[model_providers.cliproxyapi]
+name = "cliproxyapi"
+base_url = "http://127.0.0.1:8317/v1"
+wire_api = "responses"
+```
+
+auth.json:
+```json
+{
+  "OPENAI_API_KEY": "sk-dummy"
+}
+```
 
 ## 使用 Docker 运行
 
@@ -421,6 +520,12 @@ export ANTHROPIC_SMALL_FAST_MODEL=qwen3-coder-flash
 docker run --rm -p 8085:8085 -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest /CLIProxyAPI/CLIProxyAPI --login
 ```
 
+运行以下命令进行登录（Gemini Web Cookie）：
+
+```bash
+docker run -it --rm -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest /CLIProxyAPI/CLIProxyAPI --gemini-web-auth
+```
+
 运行以下命令进行登录（OpenAI OAuth，端口 1455）：
 
 ```bash
@@ -433,12 +538,90 @@ docker run --rm -p 1455:1455 -v /path/to/your/config.yaml:/CLIProxyAPI/config.ya
 docker run --rm -p 54545:54545 -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest /CLIProxyAPI/CLIProxyAPI --claude-login
 ```
 
+运行以下命令进行登录（Qwen OAuth）：
+
+```bash
+docker run -it -rm -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest /CLIProxyAPI/CLIProxyAPI --qwen-login
+```
+
+
 运行以下命令启动服务器：
 
 ```bash
 docker run --rm -p 8317:8317 -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest
 ```
 
+## 使用 Docker Compose 运行
+
+1.  克隆仓库并进入目录：
+    ```bash
+    git clone https://github.com/luispater/CLIProxyAPI.git
+    cd CLIProxyAPI
+    ```
+
+2.  准备配置文件：
+    通过复制示例文件来创建 `config.yaml` 文件，并根据您的需求进行自定义。
+    ```bash
+    cp config.example.yaml config.yaml
+    ```
+    *（Windows 用户请注意：您可以在 CMD 或 PowerShell 中使用 `copy config.example.yaml config.yaml`。）*
+
+3.  启动服务：
+    -   **适用于大多数用户（推荐）：**
+        运行以下命令，使用 Docker Hub 上的预构建镜像启动服务。服务将在后台运行。
+        ```bash
+        docker compose up -d
+        ```
+    -   **适用于进阶用户：**
+        如果您修改了源代码并需要构建新镜像，请使用交互式辅助脚本：
+        -   对于 Windows (PowerShell):
+            ```powershell
+            .\docker-build.ps1
+            ```
+        -   对于 Linux/macOS:
+            ```bash
+            bash docker-build.sh
+            ```
+        脚本将提示您选择运行方式：
+        - **选项 1：使用预构建的镜像运行 (推荐)**：从镜像仓库拉取最新的官方镜像并启动容器。这是最简单的开始方式。
+        - **选项 2：从源码构建并运行 (适用于开发者)**：从本地源代码构建镜像，将其标记为 `cli-proxy-api:local`，然后启动容器。如果您需要修改源代码，此选项很有用。
+
+4. 要在容器内运行登录命令进行身份验证：
+    - **Gemini**: 
+    ```bash
+    docker compose exec cli-proxy-api /CLIProxyAPI/CLIProxyAPI -no-browser --login
+    ```
+    - **Gemini Web**:
+    ```bash
+    docker compose exec cli-proxy-api /CLIProxyAPI/CLIProxyAPI --gemini-web-auth
+    ```
+    - **OpenAI (Codex)**:
+    ```bash
+    docker compose exec cli-proxy-api /CLIProxyAPI/CLIProxyAPI -no-browser --codex-login
+    ```
+    - **Claude**:
+    ```bash
+    docker compose exec cli-proxy-api /CLIProxyAPI/CLIProxyAPI -no-browser --claude-login
+    ```
+    - **Qwen**:
+    ```bash
+    docker compose exec cli-proxy-api /CLIProxyAPI/CLIProxyAPI -no-browser --qwen-login
+    ```
+
+5.  查看服务器日志：
+    ```bash
+    docker compose logs -f
+    ```
+
+6.  停止应用程序：
+    ```bash
+    docker compose down
+    ```
+
+## 管理 API 文档
+
+请参见 [MANAGEMENT_API_CN.md](MANAGEMENT_API_CN.md)
+
 ## 贡献
 
 欢迎贡献！请随时提交 Pull Request。

cmd/server/main.go

@@ -8,15 +8,22 @@ import (
 	"flag"
 	"fmt"
 	"os"
-	"path"
+	"path/filepath"
 	"strings"
 
-	"github.com/luispater/CLIProxyAPI/internal/cmd"
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	_ "github.com/luispater/CLIProxyAPI/internal/translator"
+	"github.com/luispater/CLIProxyAPI/v5/internal/cmd"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	_ "github.com/luispater/CLIProxyAPI/v5/internal/translator"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 )
 
+var (
+	Version   = "dev"
+	Commit    = "none"
+	BuildDate = "unknown"
+)
+
 // LogFormatter defines a custom log format for logrus.
 // This formatter adds timestamp, log level, and source location information
 // to each log entry for better debugging and monitoring.
@@ -36,7 +43,7 @@ func (m *LogFormatter) Format(entry *log.Entry) ([]byte, error) {
 	timestamp := entry.Time.Format("2006-01-02 15:04:05")
 	var newLog string
 	// Customize the log format to include timestamp, level, caller file/line, and message.
-	newLog = fmt.Sprintf("[%s] [%s] [%s:%d] %s\n", timestamp, entry.Level, path.Base(entry.Caller.File), entry.Caller.Line, entry.Message)
+	newLog = fmt.Sprintf("[%s] [%s] [%s:%d] %s\n", timestamp, entry.Level, filepath.Base(entry.Caller.File), entry.Caller.Line, entry.Message)
 
 	b.WriteString(newLog)
 	return b.Bytes(), nil
@@ -58,11 +65,14 @@ func init() {
 // It parses command-line flags, loads configuration, and starts the appropriate
 // service based on the provided flags (login, codex-login, or server mode).
 func main() {
+	log.Infof("CLIProxyAPI Version: %s, Commit: %s, BuiltAt: %s", Version, Commit, BuildDate)
+
 	// Command-line flags to control the application's behavior.
 	var login bool
 	var codexLogin bool
 	var claudeLogin bool
 	var qwenLogin bool
+	var geminiWebAuth bool
 	var noBrowser bool
 	var projectID string
 	var configPath string
@@ -72,6 +82,7 @@ func main() {
 	flag.BoolVar(&codexLogin, "codex-login", false, "Login to Codex using OAuth")
 	flag.BoolVar(&claudeLogin, "claude-login", false, "Login to Claude using OAuth")
 	flag.BoolVar(&qwenLogin, "qwen-login", false, "Login to Qwen using OAuth")
+	flag.BoolVar(&geminiWebAuth, "gemini-web-auth", false, "Auth Gemini Web using cookies")
 	flag.BoolVar(&noBrowser, "no-browser", false, "Don't open browser automatically for OAuth")
 	flag.StringVar(&projectID, "project_id", "", "Project ID (Gemini only, not required)")
 	flag.StringVar(&configPath, "config", "", "Configure File Path")
@@ -96,7 +107,7 @@ func main() {
 		if err != nil {
 			log.Fatalf("failed to get working directory: %v", err)
 		}
-		configFilePath = path.Join(wd, "config.yaml")
+		configFilePath = filepath.Join(wd, "config.yaml")
 		cfg, err = config.LoadConfig(configFilePath)
 	}
 	if err != nil {
@@ -104,11 +115,7 @@ func main() {
 	}
 
 	// Set the log level based on the configuration.
-	if cfg.Debug {
-		log.SetLevel(log.DebugLevel)
-	} else {
-		log.SetLevel(log.InfoLevel)
-	}
+	util.SetLogLevel(cfg)
 
 	// Expand the tilde (~) in the auth directory path to the user's home directory.
 	if strings.HasPrefix(cfg.AuthDir, "~") {
@@ -117,13 +124,14 @@ func main() {
 			log.Fatalf("failed to get home directory: %v", errUserHomeDir)
 		}
 		// Reconstruct the path by replacing the tilde with the user's home directory.
-		parts := strings.Split(cfg.AuthDir, string(os.PathSeparator))
-		if len(parts) > 1 {
-			parts[0] = home
-			cfg.AuthDir = path.Join(parts...)
-		} else {
-			// If the path is just "~", set it to the home directory.
+		remainder := strings.TrimPrefix(cfg.AuthDir, "~")
+		remainder = strings.TrimLeft(remainder, "/\\")
+		if remainder == "" {
 			cfg.AuthDir = home
+		} else {
+			// Normalize any slash style in the remainder so Windows paths keep nested directories.
+			normalized := strings.ReplaceAll(remainder, "\\", "/")
+			cfg.AuthDir = filepath.Join(home, filepath.FromSlash(normalized))
 		}
 	}
 
@@ -145,6 +153,8 @@ func main() {
 		cmd.DoClaudeLogin(cfg, options)
 	} else if qwenLogin {
 		cmd.DoQwenLogin(cfg, options)
+	} else if geminiWebAuth {
+		cmd.DoGeminiWebAuth(cfg)
 	} else {
 		// Start the main proxy service
 		cmd.StartService(cfg, configFilePath)

config.example.yaml

@@ -1,6 +1,17 @@
 # Server port
 port: 8317
 
+# Management API settings
+remote-management:
+  # Whether to allow remote (non-localhost) management access.
+  # When false, only localhost can access management endpoints (a key is still required).
+  allow-remote: false
+
+  # Management key. If a plaintext value is provided here, it will be hashed on startup.
+  # All management requests (even from localhost) require this key.
+  # Leave empty to disable the Management API entirely (404 for all /v0/management routes).
+  secret-key: ""
+
 # Authentication directory (supports ~ for home directory)
 auth-dir: "~/.cli-proxy-api"
 
@@ -30,6 +41,14 @@ generative-language-api-key:
   - "AIzaSy...03"
   - "AIzaSy...04"
 
+# forces the use of GPT-5 Codex model.
+force-gpt-5-codex: true
+
+# Codex API keys
+codex-api-key:
+  - api-key: "sk-atSM..."
+    base-url: "https://www.example.com" # use the custom codex API endpoint
+
 # Claude API keys
 claude-api-key:
   - api-key: "sk-atSM..." # use the official claude API key, no need to set the base url
@@ -45,4 +64,24 @@ openai-compatibility:
       - "sk-or-v1-...b781"
     models: # The models supported by the provider.
       - name: "moonshotai/kimi-k2:free" # The actual model name.
-        alias: "kimi-k2" # The alias used in the API.
+        alias: "kimi-k2" # The alias used in the API.
+
+# Gemini Web settings
+# gemini-web:
+#     # Conversation reuse: set to true to enable (default), false to disable.
+#     context: true
+#     # Maximum characters per single request to Gemini Web. Requests exceeding this
+#     # size split into chunks. Only the last chunk carries files and yields the final answer.
+#     max-chars-per-request: 1000000
+#     # Disable the short continuation hint appended to intermediate chunks
+#     # when splitting long prompts. Default is false (hint enabled by default).
+#     disable-continuation-hint: false
+#     # Background token auto-refresh interval seconds (defaults to 540 if unset or <= 0)
+#     token-refresh-seconds: 540
+#     # Code mode:
+#     #   - true: enable XML wrapping hint and attach the coding-partner Gem.
+#     #           Thought merging (<think> into visible content) applies to STREAMING only;
+#     #           non-stream responses keep reasoning/thought parts separate for clients
+#     #           that expect explicit reasoning fields.
+#     #   - false: disable XML hint and keep <think> separate
+#     code-mode: false

docker-build.ps1

@@ -0,0 +1,53 @@
+# build.ps1 - Windows PowerShell Build Script
+#
+# This script automates the process of building and running the Docker container
+# with version information dynamically injected at build time.
+
+# Stop script execution on any error
+$ErrorActionPreference = "Stop"
+
+# --- Step 1: Choose Environment ---
+Write-Host "Please select an option:"
+Write-Host "1) Run using Pre-built Image (Recommended)"
+Write-Host "2) Build from Source and Run (For Developers)"
+$choice = Read-Host -Prompt "Enter choice [1-2]"
+
+# --- Step 2: Execute based on choice ---
+switch ($choice) {
+    "1" {
+        Write-Host "--- Running with Pre-built Image ---"
+        docker compose up -d --remove-orphans --no-build
+        Write-Host "Services are starting from remote image."
+        Write-Host "Run 'docker compose logs -f' to see the logs."
+    }
+    "2" {
+        Write-Host "--- Building from Source and Running ---"
+
+        # Get Version Information
+        $VERSION = (git describe --tags --always --dirty)
+        $COMMIT  = (git rev-parse --short HEAD)
+        $BUILD_DATE = (Get-Date).ToUniversalTime().ToString("yyyy-MM-ddTHH:mm:ssZ")
+
+        Write-Host "Building with the following info:"
+        Write-Host "  Version: $VERSION"
+        Write-Host "  Commit: $COMMIT"
+        Write-Host "  Build Date: $BUILD_DATE"
+        Write-Host "----------------------------------------"
+
+        # Build and start the services with a local-only image tag
+        $env:CLI_PROXY_IMAGE = "cli-proxy-api:local"
+        
+        Write-Host "Building the Docker image..."
+        docker compose build --build-arg VERSION=$VERSION --build-arg COMMIT=$COMMIT --build-arg BUILD_DATE=$BUILD_DATE
+
+        Write-Host "Starting the services..."
+        docker compose up -d --remove-orphans --pull never
+
+        Write-Host "Build complete. Services are starting."
+        Write-Host "Run 'docker compose logs -f' to see the logs."
+    }
+    default {
+        Write-Host "Invalid choice. Please enter 1 or 2."
+        exit 1
+    }
+}

docker-build.sh

@@ -0,0 +1,58 @@
+#!/usr/bin/env bash
+#
+# build.sh - Linux/macOS Build Script
+#
+# This script automates the process of building and running the Docker container
+# with version information dynamically injected at build time.
+
+# Exit immediately if a command exits with a non-zero status.
+set -euo pipefail
+
+# --- Step 1: Choose Environment ---
+echo "Please select an option:"
+echo "1) Run using Pre-built Image (Recommended)"
+echo "2) Build from Source and Run (For Developers)"
+read -r -p "Enter choice [1-2]: " choice
+
+# --- Step 2: Execute based on choice ---
+case "$choice" in
+  1)
+    echo "--- Running with Pre-built Image ---"
+    docker compose up -d --remove-orphans --no-build
+    echo "Services are starting from remote image."
+    echo "Run 'docker compose logs -f' to see the logs."
+    ;;
+  2)
+    echo "--- Building from Source and Running ---"
+
+    # Get Version Information
+    VERSION="$(git describe --tags --always --dirty)"
+    COMMIT="$(git rev-parse --short HEAD)"
+    BUILD_DATE="$(date -u +%Y-%m-%dT%H:%M:%SZ)"
+
+    echo "Building with the following info:"
+    echo "  Version: ${VERSION}"
+    echo "  Commit: ${COMMIT}"
+    echo "  Build Date: ${BUILD_DATE}"
+    echo "----------------------------------------"
+
+    # Build and start the services with a local-only image tag
+    export CLI_PROXY_IMAGE="cli-proxy-api:local"
+    
+    echo "Building the Docker image..."
+    docker compose build \
+      --build-arg VERSION="${VERSION}" \
+      --build-arg COMMIT="${COMMIT}" \
+      --build-arg BUILD_DATE="${BUILD_DATE}"
+
+    echo "Starting the services..."
+    docker compose up -d --remove-orphans --pull never
+
+    echo "Build complete. Services are starting."
+    echo "Run 'docker compose logs -f' to see the logs."
+    ;;
+  *)
+    echo "Invalid choice. Please enter 1 or 2."
+    exit 1
+    ;;
+esac

docker-compose.yml

@@ -0,0 +1,23 @@
+services:
+  cli-proxy-api:
+    image: ${CLI_PROXY_IMAGE:-eceasy/cli-proxy-api:latest}
+    pull_policy: always
+    build:
+      context: .
+      dockerfile: Dockerfile
+      args:
+        VERSION: ${VERSION:-dev}
+        COMMIT: ${COMMIT:-none}
+        BUILD_DATE: ${BUILD_DATE:-unknown}
+    container_name: cli-proxy-api
+    ports:
+      - "8317:8317"
+      - "8085:8085"
+      - "1455:1455"
+      - "54545:54545"
+    volumes:
+      - ./config.yaml:/CLIProxyAPI/config.yaml
+      - ./auths:/root/.cli-proxy-api
+      - ./logs:/CLIProxyAPI/logs
+      - ./conv:/CLIProxyAPI/conv
+    restart: unless-stopped

go.mod

@@ -1,4 +1,4 @@
-module github.com/luispater/CLIProxyAPI
+module github.com/luispater/CLIProxyAPI/v5
 
 go 1.24
 
@@ -10,6 +10,7 @@ require (
 	github.com/skratchdot/open-golang v0.0.0-20200116055534-eef842397966
 	github.com/tidwall/gjson v1.18.0
 	github.com/tidwall/sjson v1.2.5
+	golang.org/x/crypto v0.36.0
 	golang.org/x/net v0.37.1-0.20250305215238-2914f4677317
 	golang.org/x/oauth2 v0.30.0
 	gopkg.in/yaml.v3 v3.0.1
@@ -39,7 +40,6 @@ require (
 	github.com/twitchyliquid64/golang-asm v0.15.1 // indirect
 	github.com/ugorji/go/codec v1.2.12 // indirect
 	golang.org/x/arch v0.8.0 // indirect
-	golang.org/x/crypto v0.36.0 // indirect
 	golang.org/x/sys v0.31.0 // indirect
 	golang.org/x/text v0.23.0 // indirect
 	google.golang.org/protobuf v1.34.1 // indirect

internal/api/handlers/claude/code_handlers.go

@@ -13,10 +13,11 @@ import (
 	"time"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/api/handlers"
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/api/handlers"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 	"github.com/tidwall/gjson"
 )
@@ -133,15 +134,18 @@ func (h *ClaudeCodeAPIHandler) handleStreamingResponse(c *gin.Context, rawJSON [
 		// Ensure the client's mutex is unlocked on function exit.
 		// This prevents deadlocks and ensures proper resource cleanup
 		if cliClient != nil {
-			cliClient.GetRequestMutex().Unlock()
+			if mutex := cliClient.GetRequestMutex(); mutex != nil {
+				mutex.Unlock()
+			}
 		}
 	}()
+
+	var errorResponse *interfaces.ErrorMessage
 	retryCount := 0
 	// Main client rotation loop with quota management
 	// This loop implements a sophisticated load balancing and failover mechanism
 outLoop:
 	for retryCount <= h.Cfg.RequestRetry {
-		var errorResponse *interfaces.ErrorMessage
 		cliClient, errorResponse = h.GetClient(modelName)
 		if errorResponse != nil {
 			c.Status(errorResponse.StatusCode)
@@ -182,6 +186,8 @@ outLoop:
 			// This manages various error conditions and implements retry logic
 			case errInfo, okError := <-errChan:
 				if okError {
+					errorResponse = errInfo
+					h.LoggingAPIResponseError(cliCtx, errInfo)
 					// Special handling for quota exceeded errors
 					// If configured, attempt to switch to a different project/client
 					switch errInfo.StatusCode {
@@ -191,9 +197,21 @@ outLoop:
 							continue outLoop // Restart the client selection process
 						}
 					case 403, 408, 500, 502, 503, 504:
-						log.Debugf("http status code %d, switch client", errInfo.StatusCode)
+						log.Debugf("http status code %d, switch client, %s", errInfo.StatusCode, util.HideAPIKey(cliClient.GetEmail()))
 						retryCount++
 						continue outLoop
+					case 401:
+						log.Debugf("unauthorized request, try to refresh token, %s", util.HideAPIKey(cliClient.GetEmail()))
+						err := cliClient.RefreshTokens(cliCtx)
+						if err != nil {
+							log.Debugf("refresh token failed, switch client, %s", util.HideAPIKey(cliClient.GetEmail()))
+							cliClient.SetUnavailable()
+						}
+						retryCount++
+						continue outLoop
+					case 402:
+						cliClient.SetUnavailable()
+						continue outLoop
 					default:
 						// Forward other errors directly to the client
 						c.Status(errInfo.StatusCode)
@@ -210,4 +228,12 @@ outLoop:
 			}
 		}
 	}
+
+	if errorResponse != nil {
+		c.Status(errorResponse.StatusCode)
+		_, _ = fmt.Fprint(c.Writer, errorResponse.Error.Error())
+		flusher.Flush()
+		cliCancel(errorResponse.Error)
+		return
+	}
 }

internal/api/handlers/gemini/gemini-cli_handlers.go

@@ -14,10 +14,10 @@ import (
 	"time"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/api/handlers"
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/util"
+	"github.com/luispater/CLIProxyAPI/v5/internal/api/handlers"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 	"github.com/tidwall/gjson"
 )
@@ -163,14 +163,16 @@ func (h *GeminiCLIAPIHandler) handleInternalStreamGenerateContent(c *gin.Context
 	defer func() {
 		// Ensure the client's mutex is unlocked on function exit.
 		if cliClient != nil {
-			cliClient.GetRequestMutex().Unlock()
+			if mutex := cliClient.GetRequestMutex(); mutex != nil {
+				mutex.Unlock()
+			}
 		}
 	}()
 
+	var errorResponse *interfaces.ErrorMessage
 	retryCount := 0
 outLoop:
 	for retryCount <= h.Cfg.RequestRetry {
-		var errorResponse *interfaces.ErrorMessage
 		cliClient, errorResponse = h.GetClient(modelName)
 		if errorResponse != nil {
 			c.Status(errorResponse.StatusCode)
@@ -206,6 +208,9 @@ outLoop:
 			// Handle errors from the backend.
 			case err, okError := <-errChan:
 				if okError {
+					errorResponse = err
+					h.LoggingAPIResponseError(cliCtx, err)
+
 					switch err.StatusCode {
 					case 429:
 						if h.Cfg.QuotaExceeded.SwitchProject {
@@ -216,6 +221,18 @@ outLoop:
 						log.Debugf("http status code %d, switch client", err.StatusCode)
 						retryCount++
 						continue outLoop
+					case 401:
+						log.Debugf("unauthorized request, try to refresh token, %s", util.HideAPIKey(cliClient.GetEmail()))
+						errRefreshTokens := cliClient.RefreshTokens(cliCtx)
+						if errRefreshTokens != nil {
+							log.Debugf("refresh token failed, switch client, %s", util.HideAPIKey(cliClient.GetEmail()))
+							cliClient.SetUnavailable()
+						}
+						retryCount++
+						continue outLoop
+					case 402:
+						cliClient.SetUnavailable()
+						continue outLoop
 					default:
 						// Forward other errors directly to the client
 						c.Status(err.StatusCode)
@@ -230,6 +247,13 @@ outLoop:
 			}
 		}
 	}
+	if errorResponse != nil {
+		c.Status(errorResponse.StatusCode)
+		_, _ = fmt.Fprint(c.Writer, errorResponse.Error.Error())
+		flusher.Flush()
+		cliCancel(errorResponse.Error)
+		return
+	}
 }
 
 // handleInternalGenerateContent handles non-streaming content generation requests.
@@ -244,13 +268,15 @@ func (h *GeminiCLIAPIHandler) handleInternalGenerateContent(c *gin.Context, rawJ
 	var cliClient interfaces.Client
 	defer func() {
 		if cliClient != nil {
-			cliClient.GetRequestMutex().Unlock()
+			if mutex := cliClient.GetRequestMutex(); mutex != nil {
+				mutex.Unlock()
+			}
 		}
 	}()
 
+	var errorResponse *interfaces.ErrorMessage
 	retryCount := 0
 	for retryCount <= h.Cfg.RequestRetry {
-		var errorResponse *interfaces.ErrorMessage
 		cliClient, errorResponse = h.GetClient(modelName)
 		if errorResponse != nil {
 			c.Status(errorResponse.StatusCode)
@@ -261,6 +287,9 @@ func (h *GeminiCLIAPIHandler) handleInternalGenerateContent(c *gin.Context, rawJ
 
 		resp, err := cliClient.SendRawMessage(cliCtx, modelName, rawJSON, "")
 		if err != nil {
+			errorResponse = err
+			h.LoggingAPIResponseError(cliCtx, err)
+
 			switch err.StatusCode {
 			case 429:
 				if h.Cfg.QuotaExceeded.SwitchProject {
@@ -271,6 +300,18 @@ func (h *GeminiCLIAPIHandler) handleInternalGenerateContent(c *gin.Context, rawJ
 				log.Debugf("http status code %d, switch client", err.StatusCode)
 				retryCount++
 				continue
+			case 401:
+				log.Debugf("unauthorized request, try to refresh token, %s", util.HideAPIKey(cliClient.GetEmail()))
+				errRefreshTokens := cliClient.RefreshTokens(cliCtx)
+				if errRefreshTokens != nil {
+					log.Debugf("refresh token failed, switch client, %s", util.HideAPIKey(cliClient.GetEmail()))
+					cliClient.SetUnavailable()
+				}
+				retryCount++
+				continue
+			case 402:
+				cliClient.SetUnavailable()
+				continue
 			default:
 				// Forward other errors directly to the client
 				c.Status(err.StatusCode)
@@ -280,8 +321,15 @@ func (h *GeminiCLIAPIHandler) handleInternalGenerateContent(c *gin.Context, rawJ
 			break
 		} else {
 			_, _ = c.Writer.Write(resp)
-			cliCancel(resp)
+			cliCancel()
 			break
 		}
 	}
+	if errorResponse != nil {
+		c.Status(errorResponse.StatusCode)
+		_, _ = c.Writer.Write([]byte(errorResponse.Error.Error()))
+		cliCancel(errorResponse.Error)
+		return
+	}
+
 }

internal/api/handlers/gemini/gemini_handlers.go

@@ -13,10 +13,11 @@ import (
 	"time"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/api/handlers"
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/api/handlers"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 )
 
@@ -214,14 +215,16 @@ func (h *GeminiAPIHandler) handleStreamGenerateContent(c *gin.Context, modelName
 	defer func() {
 		// Ensure the client's mutex is unlocked on function exit.
 		if cliClient != nil {
-			cliClient.GetRequestMutex().Unlock()
+			if mutex := cliClient.GetRequestMutex(); mutex != nil {
+				mutex.Unlock()
+			}
 		}
 	}()
 
+	var errorResponse *interfaces.ErrorMessage
 	retryCount := 0
 outLoop:
 	for retryCount <= h.Cfg.RequestRetry {
-		var errorResponse *interfaces.ErrorMessage
 		cliClient, errorResponse = h.GetClient(modelName)
 		if errorResponse != nil {
 			c.Status(errorResponse.StatusCode)
@@ -260,6 +263,9 @@ outLoop:
 			// Handle errors from the backend.
 			case err, okError := <-errChan:
 				if okError {
+					errorResponse = err
+					h.LoggingAPIResponseError(cliCtx, err)
+
 					switch err.StatusCode {
 					case 429:
 						if h.Cfg.QuotaExceeded.SwitchProject {
@@ -270,6 +276,18 @@ outLoop:
 						log.Debugf("http status code %d, switch client", err.StatusCode)
 						retryCount++
 						continue outLoop
+					case 401:
+						log.Debugf("unauthorized request, try to refresh token, %s", util.HideAPIKey(cliClient.GetEmail()))
+						errRefreshTokens := cliClient.RefreshTokens(cliCtx)
+						if errRefreshTokens != nil {
+							log.Debugf("refresh token failed, switch client, %s", util.HideAPIKey(cliClient.GetEmail()))
+							cliClient.SetUnavailable()
+						}
+						retryCount++
+						continue outLoop
+					case 402:
+						cliClient.SetUnavailable()
+						continue outLoop
 					default:
 						// Forward other errors directly to the client
 						c.Status(err.StatusCode)
@@ -284,6 +302,13 @@ outLoop:
 			}
 		}
 	}
+	if errorResponse != nil {
+		c.Status(errorResponse.StatusCode)
+		_, _ = fmt.Fprint(c.Writer, errorResponse.Error.Error())
+		flusher.Flush()
+		cliCancel(errorResponse.Error)
+		return
+	}
 }
 
 // handleCountTokens handles token counting requests for Gemini models.
@@ -303,7 +328,9 @@ func (h *GeminiAPIHandler) handleCountTokens(c *gin.Context, modelName string, r
 	var cliClient interfaces.Client
 	defer func() {
 		if cliClient != nil {
-			cliClient.GetRequestMutex().Unlock()
+			if mutex := cliClient.GetRequestMutex(); mutex != nil {
+				mutex.Unlock()
+			}
 		}
 	}()
 
@@ -354,13 +381,15 @@ func (h *GeminiAPIHandler) handleGenerateContent(c *gin.Context, modelName strin
 	var cliClient interfaces.Client
 	defer func() {
 		if cliClient != nil {
-			cliClient.GetRequestMutex().Unlock()
+			if mutex := cliClient.GetRequestMutex(); mutex != nil {
+				mutex.Unlock()
+			}
 		}
 	}()
 
+	var errorResponse *interfaces.ErrorMessage
 	retryCount := 0
 	for retryCount <= h.Cfg.RequestRetry {
-		var errorResponse *interfaces.ErrorMessage
 		cliClient, errorResponse = h.GetClient(modelName)
 		if errorResponse != nil {
 			c.Status(errorResponse.StatusCode)
@@ -371,6 +400,9 @@ func (h *GeminiAPIHandler) handleGenerateContent(c *gin.Context, modelName strin
 
 		resp, err := cliClient.SendRawMessage(cliCtx, modelName, rawJSON, alt)
 		if err != nil {
+			errorResponse = err
+			h.LoggingAPIResponseError(cliCtx, err)
+
 			switch err.StatusCode {
 			case 429:
 				if h.Cfg.QuotaExceeded.SwitchProject {
@@ -381,6 +413,18 @@ func (h *GeminiAPIHandler) handleGenerateContent(c *gin.Context, modelName strin
 				log.Debugf("http status code %d, switch client", err.StatusCode)
 				retryCount++
 				continue
+			case 401:
+				log.Debugf("unauthorized request, try to refresh token, %s", util.HideAPIKey(cliClient.GetEmail()))
+				errRefreshTokens := cliClient.RefreshTokens(cliCtx)
+				if errRefreshTokens != nil {
+					log.Debugf("refresh token failed, switch client, %s", util.HideAPIKey(cliClient.GetEmail()))
+					cliClient.SetUnavailable()
+				}
+				retryCount++
+				continue
+			case 402:
+				cliClient.SetUnavailable()
+				continue
 			default:
 				// Forward other errors directly to the client
 				c.Status(err.StatusCode)
@@ -390,8 +434,14 @@ func (h *GeminiAPIHandler) handleGenerateContent(c *gin.Context, modelName strin
 			break
 		} else {
 			_, _ = c.Writer.Write(resp)
-			cliCancel(resp)
+			cliCancel()
 			break
 		}
 	}
+	if errorResponse != nil {
+		c.Status(errorResponse.StatusCode)
+		_, _ = c.Writer.Write([]byte(errorResponse.Error.Error()))
+		cliCancel(errorResponse.Error)
+		return
+	}
 }

internal/api/handlers/handlers.go

@@ -8,11 +8,9 @@ import (
 	"sync"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/client"
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/util"
-	log "github.com/sirupsen/logrus"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	"golang.org/x/net/context"
 )
 
@@ -97,23 +95,24 @@ func (h *BaseAPIHandler) UpdateClients(clients []interfaces.Client, cfg *config.
 func (h *BaseAPIHandler) GetClient(modelName string, isGenerateContent ...bool) (interfaces.Client, *interfaces.ErrorMessage) {
 	clients := make([]interfaces.Client, 0)
 	for i := 0; i < len(h.CliClients); i++ {
-		if h.CliClients[i].CanProvideModel(modelName) {
+		if h.CliClients[i].CanProvideModel(modelName) && h.CliClients[i].IsAvailable() && !h.CliClients[i].IsModelQuotaExceeded(modelName) {
 			clients = append(clients, h.CliClients[i])
 		}
 	}
 
+	// Lock the mutex to update the last used client index
+	h.Mutex.Lock()
 	if _, hasKey := h.LastUsedClientIndex[modelName]; !hasKey {
 		h.LastUsedClientIndex[modelName] = 0
 	}
 
 	if len(clients) == 0 {
+		h.Mutex.Unlock()
 		return nil, &interfaces.ErrorMessage{StatusCode: 500, Error: fmt.Errorf("no clients available")}
 	}
 
 	var cliClient interfaces.Client
 
-	// Lock the mutex to update the last used client index
-	h.Mutex.Lock()
 	startIndex := h.LastUsedClientIndex[modelName]
 	if (len(isGenerateContent) > 0 && isGenerateContent[0]) || len(isGenerateContent) == 0 {
 		currentIndex := (startIndex + 1) % len(clients)
@@ -125,24 +124,6 @@ func (h *BaseAPIHandler) GetClient(modelName string, isGenerateContent ...bool)
 	reorderedClients := make([]interfaces.Client, 0)
 	for i := 0; i < len(clients); i++ {
 		cliClient = clients[(startIndex+1+i)%len(clients)]
-		if cliClient.IsModelQuotaExceeded(modelName) {
-			if cliClient.Provider() == "gemini-cli" {
-				log.Debugf("Gemini Model %s is quota exceeded for account %s, project id: %s", modelName, cliClient.GetEmail(), cliClient.(*client.GeminiCLIClient).GetProjectID())
-			} else if cliClient.Provider() == "gemini" {
-				log.Debugf("Gemini Model %s is quota exceeded for account %s", modelName, cliClient.GetEmail())
-			} else if cliClient.Provider() == "codex" {
-				log.Debugf("Codex Model %s is quota exceeded for account %s", modelName, cliClient.GetEmail())
-			} else if cliClient.Provider() == "claude" {
-				log.Debugf("Claude Model %s is quota exceeded for account %s", modelName, cliClient.GetEmail())
-			} else if cliClient.Provider() == "qwen" {
-				log.Debugf("Qwen Model %s is quota exceeded for account %s", modelName, cliClient.GetEmail())
-			} else if cliClient.Type() == "openai-compatibility" {
-				log.Debugf("OpenAI Compatibility Model %s is quota exceeded for provider %s", modelName, cliClient.Provider())
-			}
-			cliClient = nil
-			continue
-
-		}
 		reorderedClients = append(reorderedClients, cliClient)
 	}
 
@@ -157,14 +138,20 @@ func (h *BaseAPIHandler) GetClient(modelName string, isGenerateContent ...bool)
 	locked := false
 	for i := 0; i < len(reorderedClients); i++ {
 		cliClient = reorderedClients[i]
-		if cliClient.GetRequestMutex().TryLock() {
+		if mutex := cliClient.GetRequestMutex(); mutex != nil {
+			if mutex.TryLock() {
+				locked = true
+				break
+			}
+		} else {
 			locked = true
-			break
 		}
 	}
 	if !locked {
 		cliClient = clients[0]
-		cliClient.GetRequestMutex().Lock()
+		if mutex := cliClient.GetRequestMutex(); mutex != nil {
+			mutex.Lock()
+		}
 	}
 
 	return cliClient, nil
@@ -228,6 +215,22 @@ func (h *BaseAPIHandler) GetContextWithCancel(handler interfaces.APIHandler, c *
 	}
 }
 
+func (h *BaseAPIHandler) LoggingAPIResponseError(ctx context.Context, err *interfaces.ErrorMessage) {
+	if h.Cfg.RequestLog {
+		if ginContext, ok := ctx.Value("gin").(*gin.Context); ok {
+			if apiResponseErrors, isExist := ginContext.Get("API_RESPONSE_ERROR"); isExist {
+				if slicesAPIResponseError, isOk := apiResponseErrors.([]*interfaces.ErrorMessage); isOk {
+					slicesAPIResponseError = append(slicesAPIResponseError, err)
+					ginContext.Set("API_RESPONSE_ERROR", slicesAPIResponseError)
+				}
+			} else {
+				// Create new response data entry
+				ginContext.Set("API_RESPONSE_ERROR", []*interfaces.ErrorMessage{err})
+			}
+		}
+	}
+}
+
 // APIHandlerCancelFunc is a function type for canceling an API handler's context.
 // It can optionally accept parameters, which are used for logging the response.
 type APIHandlerCancelFunc func(params ...interface{})

internal/api/handlers/management/auth_files.go

@@ -0,0 +1,744 @@
+package management
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"os"
+	"path/filepath"
+	"strings"
+	"time"
+
+	"github.com/gin-gonic/gin"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/claude"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/codex"
+	geminiAuth "github.com/luispater/CLIProxyAPI/v5/internal/auth/gemini"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/qwen"
+	"github.com/luispater/CLIProxyAPI/v5/internal/client"
+	"github.com/luispater/CLIProxyAPI/v5/internal/misc"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
+	log "github.com/sirupsen/logrus"
+	"github.com/tidwall/gjson"
+	"golang.org/x/oauth2"
+	"golang.org/x/oauth2/google"
+)
+
+var (
+	oauthStatus = make(map[string]string)
+)
+
+// List auth files
+func (h *Handler) ListAuthFiles(c *gin.Context) {
+	entries, err := os.ReadDir(h.cfg.AuthDir)
+	if err != nil {
+		c.JSON(500, gin.H{"error": fmt.Sprintf("failed to read auth dir: %v", err)})
+		return
+	}
+	files := make([]gin.H, 0)
+	for _, e := range entries {
+		if e.IsDir() {
+			continue
+		}
+		name := e.Name()
+		if !strings.HasSuffix(strings.ToLower(name), ".json") {
+			continue
+		}
+		if info, errInfo := e.Info(); errInfo == nil {
+			fileData := gin.H{"name": name, "size": info.Size(), "modtime": info.ModTime()}
+
+			// Read file to get type field
+			full := filepath.Join(h.cfg.AuthDir, name)
+			if data, errRead := os.ReadFile(full); errRead == nil {
+				typeValue := gjson.GetBytes(data, "type").String()
+				fileData["type"] = typeValue
+			}
+
+			files = append(files, fileData)
+		}
+	}
+	c.JSON(200, gin.H{"files": files})
+}
+
+// Download single auth file by name
+func (h *Handler) DownloadAuthFile(c *gin.Context) {
+	name := c.Query("name")
+	if name == "" || strings.Contains(name, string(os.PathSeparator)) {
+		c.JSON(400, gin.H{"error": "invalid name"})
+		return
+	}
+	if !strings.HasSuffix(strings.ToLower(name), ".json") {
+		c.JSON(400, gin.H{"error": "name must end with .json"})
+		return
+	}
+	full := filepath.Join(h.cfg.AuthDir, name)
+	data, err := os.ReadFile(full)
+	if err != nil {
+		if os.IsNotExist(err) {
+			c.JSON(404, gin.H{"error": "file not found"})
+		} else {
+			c.JSON(500, gin.H{"error": fmt.Sprintf("failed to read file: %v", err)})
+		}
+		return
+	}
+	c.Header("Content-Disposition", fmt.Sprintf("attachment; filename=\"%s\"", name))
+	c.Data(200, "application/json", data)
+}
+
+// Upload auth file: multipart or raw JSON with ?name=
+func (h *Handler) UploadAuthFile(c *gin.Context) {
+	if file, err := c.FormFile("file"); err == nil && file != nil {
+		name := filepath.Base(file.Filename)
+		if !strings.HasSuffix(strings.ToLower(name), ".json") {
+			c.JSON(400, gin.H{"error": "file must be .json"})
+			return
+		}
+		dst := filepath.Join(h.cfg.AuthDir, name)
+		if errSave := c.SaveUploadedFile(file, dst); errSave != nil {
+			c.JSON(500, gin.H{"error": fmt.Sprintf("failed to save file: %v", errSave)})
+			return
+		}
+		c.JSON(200, gin.H{"status": "ok"})
+		return
+	}
+	name := c.Query("name")
+	if name == "" || strings.Contains(name, string(os.PathSeparator)) {
+		c.JSON(400, gin.H{"error": "invalid name"})
+		return
+	}
+	if !strings.HasSuffix(strings.ToLower(name), ".json") {
+		c.JSON(400, gin.H{"error": "name must end with .json"})
+		return
+	}
+	data, err := io.ReadAll(c.Request.Body)
+	if err != nil {
+		c.JSON(400, gin.H{"error": "failed to read body"})
+		return
+	}
+	dst := filepath.Join(h.cfg.AuthDir, filepath.Base(name))
+	if errWrite := os.WriteFile(dst, data, 0o600); errWrite != nil {
+		c.JSON(500, gin.H{"error": fmt.Sprintf("failed to write file: %v", errWrite)})
+		return
+	}
+	c.JSON(200, gin.H{"status": "ok"})
+}
+
+// Delete auth files: single by name or all
+func (h *Handler) DeleteAuthFile(c *gin.Context) {
+	if all := c.Query("all"); all == "true" || all == "1" || all == "*" {
+		entries, err := os.ReadDir(h.cfg.AuthDir)
+		if err != nil {
+			c.JSON(500, gin.H{"error": fmt.Sprintf("failed to read auth dir: %v", err)})
+			return
+		}
+		deleted := 0
+		for _, e := range entries {
+			if e.IsDir() {
+				continue
+			}
+			name := e.Name()
+			if !strings.HasSuffix(strings.ToLower(name), ".json") {
+				continue
+			}
+			full := filepath.Join(h.cfg.AuthDir, name)
+			if err = os.Remove(full); err == nil {
+				deleted++
+			}
+		}
+		c.JSON(200, gin.H{"status": "ok", "deleted": deleted})
+		return
+	}
+	name := c.Query("name")
+	if name == "" || strings.Contains(name, string(os.PathSeparator)) {
+		c.JSON(400, gin.H{"error": "invalid name"})
+		return
+	}
+	full := filepath.Join(h.cfg.AuthDir, filepath.Base(name))
+	if err := os.Remove(full); err != nil {
+		if os.IsNotExist(err) {
+			c.JSON(404, gin.H{"error": "file not found"})
+		} else {
+			c.JSON(500, gin.H{"error": fmt.Sprintf("failed to remove file: %v", err)})
+		}
+		return
+	}
+	c.JSON(200, gin.H{"status": "ok"})
+}
+
+func (h *Handler) RequestAnthropicToken(c *gin.Context) {
+	ctx := context.Background()
+
+	log.Info("Initializing Claude authentication...")
+
+	// Generate PKCE codes
+	pkceCodes, err := claude.GeneratePKCECodes()
+	if err != nil {
+		log.Fatalf("Failed to generate PKCE codes: %v", err)
+		return
+	}
+
+	// Generate random state parameter
+	state, err := misc.GenerateRandomState()
+	if err != nil {
+		log.Fatalf("Failed to generate state parameter: %v", err)
+		return
+	}
+
+	// Initialize Claude auth service
+	anthropicAuth := claude.NewClaudeAuth(h.cfg)
+
+	// Generate authorization URL (then override redirect_uri to reuse server port)
+	authURL, state, err := anthropicAuth.GenerateAuthURL(state, pkceCodes)
+	if err != nil {
+		log.Fatalf("Failed to generate authorization URL: %v", err)
+		return
+	}
+	// Override redirect_uri in authorization URL to current server port
+
+	go func() {
+		// Helper: wait for callback file
+		waitFile := filepath.Join(h.cfg.AuthDir, fmt.Sprintf(".oauth-anthropic-%s.oauth", state))
+		waitForFile := func(path string, timeout time.Duration) (map[string]string, error) {
+			deadline := time.Now().Add(timeout)
+			for {
+				if time.Now().After(deadline) {
+					oauthStatus[state] = "Timeout waiting for OAuth callback"
+					return nil, fmt.Errorf("timeout waiting for OAuth callback")
+				}
+				data, errRead := os.ReadFile(path)
+				if errRead == nil {
+					var m map[string]string
+					_ = json.Unmarshal(data, &m)
+					_ = os.Remove(path)
+					return m, nil
+				}
+				time.Sleep(500 * time.Millisecond)
+			}
+		}
+
+		log.Info("Waiting for authentication callback...")
+		// Wait up to 5 minutes
+		resultMap, errWait := waitForFile(waitFile, 5*time.Minute)
+		if errWait != nil {
+			authErr := claude.NewAuthenticationError(claude.ErrCallbackTimeout, errWait)
+			log.Error(claude.GetUserFriendlyMessage(authErr))
+			return
+		}
+		if errStr := resultMap["error"]; errStr != "" {
+			oauthErr := claude.NewOAuthError(errStr, "", http.StatusBadRequest)
+			log.Error(claude.GetUserFriendlyMessage(oauthErr))
+			oauthStatus[state] = "Bad request"
+			return
+		}
+		if resultMap["state"] != state {
+			authErr := claude.NewAuthenticationError(claude.ErrInvalidState, fmt.Errorf("expected %s, got %s", state, resultMap["state"]))
+			log.Error(claude.GetUserFriendlyMessage(authErr))
+			oauthStatus[state] = "State code error"
+			return
+		}
+
+		// Parse code (Claude may append state after '#')
+		rawCode := resultMap["code"]
+		code := strings.Split(rawCode, "#")[0]
+
+		// Exchange code for tokens (replicate logic using updated redirect_uri)
+		// Extract client_id from the modified auth URL
+		clientID := ""
+		if u2, errP := url.Parse(authURL); errP == nil {
+			clientID = u2.Query().Get("client_id")
+		}
+		// Build request
+		bodyMap := map[string]any{
+			"code":          code,
+			"state":         state,
+			"grant_type":    "authorization_code",
+			"client_id":     clientID,
+			"redirect_uri":  "http://localhost:54545/callback",
+			"code_verifier": pkceCodes.CodeVerifier,
+		}
+		bodyJSON, _ := json.Marshal(bodyMap)
+
+		httpClient := util.SetProxy(h.cfg, &http.Client{})
+		req, _ := http.NewRequestWithContext(ctx, "POST", "https://console.anthropic.com/v1/oauth/token", strings.NewReader(string(bodyJSON)))
+		req.Header.Set("Content-Type", "application/json")
+		req.Header.Set("Accept", "application/json")
+		resp, errDo := httpClient.Do(req)
+		if errDo != nil {
+			authErr := claude.NewAuthenticationError(claude.ErrCodeExchangeFailed, errDo)
+			log.Errorf("Failed to exchange authorization code for tokens: %v", authErr)
+			oauthStatus[state] = "Failed to exchange authorization code for tokens"
+			return
+		}
+		defer func() {
+			if errClose := resp.Body.Close(); errClose != nil {
+				log.Errorf("failed to close response body: %v", errClose)
+			}
+		}()
+		respBody, _ := io.ReadAll(resp.Body)
+		if resp.StatusCode != http.StatusOK {
+			log.Errorf("token exchange failed with status %d: %s", resp.StatusCode, string(respBody))
+			oauthStatus[state] = fmt.Sprintf("token exchange failed with status %d", resp.StatusCode)
+			return
+		}
+		var tResp struct {
+			AccessToken  string `json:"access_token"`
+			RefreshToken string `json:"refresh_token"`
+			ExpiresIn    int    `json:"expires_in"`
+			Account      struct {
+				EmailAddress string `json:"email_address"`
+			} `json:"account"`
+		}
+		if errU := json.Unmarshal(respBody, &tResp); errU != nil {
+			log.Errorf("failed to parse token response: %v", errU)
+			oauthStatus[state] = "Failed to parse token response"
+			return
+		}
+		bundle := &claude.ClaudeAuthBundle{
+			TokenData: claude.ClaudeTokenData{
+				AccessToken:  tResp.AccessToken,
+				RefreshToken: tResp.RefreshToken,
+				Email:        tResp.Account.EmailAddress,
+				Expire:       time.Now().Add(time.Duration(tResp.ExpiresIn) * time.Second).Format(time.RFC3339),
+			},
+			LastRefresh: time.Now().Format(time.RFC3339),
+		}
+
+		// Create token storage
+		tokenStorage := anthropicAuth.CreateTokenStorage(bundle)
+		// Initialize Claude client
+		anthropicClient := client.NewClaudeClient(h.cfg, tokenStorage)
+		// Save token storage
+		if errSave := anthropicClient.SaveTokenToFile(); errSave != nil {
+			log.Fatalf("Failed to save authentication tokens: %v", errSave)
+			oauthStatus[state] = "Failed to save authentication tokens"
+			return
+		}
+
+		log.Info("Authentication successful!")
+		if bundle.APIKey != "" {
+			log.Info("API key obtained and saved")
+		}
+		log.Info("You can now use Claude services through this CLI")
+		delete(oauthStatus, state)
+	}()
+
+	oauthStatus[state] = ""
+	c.JSON(200, gin.H{"status": "ok", "url": authURL, "state": state})
+}
+
+func (h *Handler) RequestGeminiCLIToken(c *gin.Context) {
+	ctx := context.Background()
+
+	// Optional project ID from query
+	projectID := c.Query("project_id")
+
+	log.Info("Initializing Google authentication...")
+
+	// OAuth2 configuration (mirrors internal/auth/gemini)
+	conf := &oauth2.Config{
+		ClientID:     "681255809395-oo8ft2oprdrnp9e3aqf6av3hmdib135j.apps.googleusercontent.com",
+		ClientSecret: "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",
+		RedirectURL:  "http://localhost:8085/oauth2callback",
+		Scopes: []string{
+			"https://www.googleapis.com/auth/cloud-platform",
+			"https://www.googleapis.com/auth/userinfo.email",
+			"https://www.googleapis.com/auth/userinfo.profile",
+		},
+		Endpoint: google.Endpoint,
+	}
+
+	// Build authorization URL and return it immediately
+	state := fmt.Sprintf("gem-%d", time.Now().UnixNano())
+	authURL := conf.AuthCodeURL(state, oauth2.AccessTypeOffline, oauth2.SetAuthURLParam("prompt", "consent"))
+
+	go func() {
+		// Wait for callback file written by server route
+		waitFile := filepath.Join(h.cfg.AuthDir, fmt.Sprintf(".oauth-gemini-%s.oauth", state))
+		log.Info("Waiting for authentication callback...")
+		deadline := time.Now().Add(5 * time.Minute)
+		var authCode string
+		for {
+			if time.Now().After(deadline) {
+				log.Error("oauth flow timed out")
+				oauthStatus[state] = "OAuth flow timed out"
+				return
+			}
+			if data, errR := os.ReadFile(waitFile); errR == nil {
+				var m map[string]string
+				_ = json.Unmarshal(data, &m)
+				_ = os.Remove(waitFile)
+				if errStr := m["error"]; errStr != "" {
+					log.Errorf("Authentication failed: %s", errStr)
+					oauthStatus[state] = "Authentication failed"
+					return
+				}
+				authCode = m["code"]
+				if authCode == "" {
+					log.Errorf("Authentication failed: code not found")
+					oauthStatus[state] = "Authentication failed: code not found"
+					return
+				}
+				break
+			}
+			time.Sleep(500 * time.Millisecond)
+		}
+
+		// Exchange authorization code for token
+		token, err := conf.Exchange(ctx, authCode)
+		if err != nil {
+			log.Errorf("Failed to exchange token: %v", err)
+			oauthStatus[state] = "Failed to exchange token"
+			return
+		}
+
+		// Create token storage (mirrors internal/auth/gemini createTokenStorage)
+		httpClient := conf.Client(ctx, token)
+		req, errNewRequest := http.NewRequestWithContext(ctx, "GET", "https://www.googleapis.com/oauth2/v1/userinfo?alt=json", nil)
+		if errNewRequest != nil {
+			log.Errorf("Could not get user info: %v", errNewRequest)
+			oauthStatus[state] = "Could not get user info"
+			return
+		}
+		req.Header.Set("Content-Type", "application/json")
+		req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", token.AccessToken))
+
+		resp, errDo := httpClient.Do(req)
+		if errDo != nil {
+			log.Errorf("Failed to execute request: %v", errDo)
+			oauthStatus[state] = "Failed to execute request"
+			return
+		}
+		defer func() {
+			if errClose := resp.Body.Close(); errClose != nil {
+				log.Printf("warn: failed to close response body: %v", errClose)
+			}
+		}()
+
+		bodyBytes, _ := io.ReadAll(resp.Body)
+		if resp.StatusCode < 200 || resp.StatusCode >= 300 {
+			log.Errorf("Get user info request failed with status %d: %s", resp.StatusCode, string(bodyBytes))
+			oauthStatus[state] = fmt.Sprintf("Get user info request failed with status %d", resp.StatusCode)
+			return
+		}
+
+		email := gjson.GetBytes(bodyBytes, "email").String()
+		if email != "" {
+			log.Infof("Authenticated user email: %s", email)
+		} else {
+			log.Info("Failed to get user email from token")
+			oauthStatus[state] = "Failed to get user email from token"
+		}
+
+		// Marshal/unmarshal oauth2.Token to generic map and enrich fields
+		var ifToken map[string]any
+		jsonData, _ := json.Marshal(token)
+		if errUnmarshal := json.Unmarshal(jsonData, &ifToken); errUnmarshal != nil {
+			log.Errorf("Failed to unmarshal token: %v", errUnmarshal)
+			oauthStatus[state] = "Failed to unmarshal token"
+			return
+		}
+
+		ifToken["token_uri"] = "https://oauth2.googleapis.com/token"
+		ifToken["client_id"] = "681255809395-oo8ft2oprdrnp9e3aqf6av3hmdib135j.apps.googleusercontent.com"
+		ifToken["client_secret"] = "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"
+		ifToken["scopes"] = []string{
+			"https://www.googleapis.com/auth/cloud-platform",
+			"https://www.googleapis.com/auth/userinfo.email",
+			"https://www.googleapis.com/auth/userinfo.profile",
+		}
+		ifToken["universe_domain"] = "googleapis.com"
+
+		ts := geminiAuth.GeminiTokenStorage{
+			Token:     ifToken,
+			ProjectID: projectID,
+			Email:     email,
+		}
+
+		// Initialize authenticated HTTP client via GeminiAuth to honor proxy settings
+		gemAuth := geminiAuth.NewGeminiAuth()
+		httpClient2, errGetClient := gemAuth.GetAuthenticatedClient(ctx, &ts, h.cfg, true)
+		if errGetClient != nil {
+			log.Fatalf("failed to get authenticated client: %v", errGetClient)
+			oauthStatus[state] = "Failed to get authenticated client"
+			return
+		}
+		log.Info("Authentication successful.")
+
+		// Initialize the API client
+		cliClient := client.NewGeminiCLIClient(httpClient2, &ts, h.cfg)
+
+		// Perform the user setup process (migrated from DoLogin)
+		if err = cliClient.SetupUser(ctx, ts.Email, projectID); err != nil {
+			if err.Error() == "failed to start user onboarding, need define a project id" {
+				log.Error("Failed to start user onboarding: A project ID is required.")
+				oauthStatus[state] = "Failed to start user onboarding: A project ID is required"
+				project, errGetProjectList := cliClient.GetProjectList(ctx)
+				if errGetProjectList != nil {
+					log.Fatalf("Failed to get project list: %v", err)
+					oauthStatus[state] = "Failed to get project list"
+				} else {
+					log.Infof("Your account %s needs to specify a project ID.", ts.Email)
+					log.Info("========================================================================")
+					for _, p := range project.Projects {
+						log.Infof("Project ID: %s", p.ProjectID)
+						log.Infof("Project Name: %s", p.Name)
+						log.Info("------------------------------------------------------------------------")
+					}
+					log.Infof("Please run this command to login again with a specific project:\n\n%s --login --project_id <project_id>\n", os.Args[0])
+				}
+			} else {
+				log.Fatalf("Failed to complete user setup: %v", err)
+				oauthStatus[state] = "Failed to complete user setup"
+			}
+			return
+		}
+
+		// Post-setup checks and token persistence
+		auto := projectID == ""
+		cliClient.SetIsAuto(auto)
+		if !cliClient.IsChecked() && !cliClient.IsAuto() {
+			isChecked, checkErr := cliClient.CheckCloudAPIIsEnabled()
+			if checkErr != nil {
+				log.Fatalf("Failed to check if Cloud AI API is enabled: %v", checkErr)
+				oauthStatus[state] = "Failed to check if Cloud AI API is enabled"
+				return
+			}
+			cliClient.SetIsChecked(isChecked)
+			if !isChecked {
+				log.Fatal("Failed to check if Cloud AI API is enabled. If you encounter an error message, please create an issue.")
+				oauthStatus[state] = "Failed to check if Cloud AI API is enabled"
+				return
+			}
+		}
+
+		if err = cliClient.SaveTokenToFile(); err != nil {
+			log.Fatalf("Failed to save token to file: %v", err)
+			oauthStatus[state] = "Failed to save token to file"
+			return
+		}
+
+		delete(oauthStatus, state)
+		log.Info("You can now use Gemini CLI services through this CLI")
+	}()
+
+	oauthStatus[state] = ""
+	c.JSON(200, gin.H{"status": "ok", "url": authURL, "state": state})
+}
+
+func (h *Handler) RequestCodexToken(c *gin.Context) {
+	ctx := context.Background()
+
+	log.Info("Initializing Codex authentication...")
+
+	// Generate PKCE codes
+	pkceCodes, err := codex.GeneratePKCECodes()
+	if err != nil {
+		log.Fatalf("Failed to generate PKCE codes: %v", err)
+		return
+	}
+
+	// Generate random state parameter
+	state, err := misc.GenerateRandomState()
+	if err != nil {
+		log.Fatalf("Failed to generate state parameter: %v", err)
+		return
+	}
+
+	// Initialize Codex auth service
+	openaiAuth := codex.NewCodexAuth(h.cfg)
+
+	// Generate authorization URL
+	authURL, err := openaiAuth.GenerateAuthURL(state, pkceCodes)
+	if err != nil {
+		log.Fatalf("Failed to generate authorization URL: %v", err)
+		return
+	}
+
+	go func() {
+		// Wait for callback file
+		waitFile := filepath.Join(h.cfg.AuthDir, fmt.Sprintf(".oauth-codex-%s.oauth", state))
+		deadline := time.Now().Add(5 * time.Minute)
+		var code string
+		for {
+			if time.Now().After(deadline) {
+				authErr := codex.NewAuthenticationError(codex.ErrCallbackTimeout, fmt.Errorf("timeout waiting for OAuth callback"))
+				log.Error(codex.GetUserFriendlyMessage(authErr))
+				oauthStatus[state] = "Timeout waiting for OAuth callback"
+				return
+			}
+			if data, errR := os.ReadFile(waitFile); errR == nil {
+				var m map[string]string
+				_ = json.Unmarshal(data, &m)
+				_ = os.Remove(waitFile)
+				if errStr := m["error"]; errStr != "" {
+					oauthErr := codex.NewOAuthError(errStr, "", http.StatusBadRequest)
+					log.Error(codex.GetUserFriendlyMessage(oauthErr))
+					oauthStatus[state] = "Bad Request"
+					return
+				}
+				if m["state"] != state {
+					authErr := codex.NewAuthenticationError(codex.ErrInvalidState, fmt.Errorf("expected %s, got %s", state, m["state"]))
+					oauthStatus[state] = "State code error"
+					log.Error(codex.GetUserFriendlyMessage(authErr))
+					return
+				}
+				code = m["code"]
+				break
+			}
+			time.Sleep(500 * time.Millisecond)
+		}
+
+		log.Debug("Authorization code received, exchanging for tokens...")
+		// Extract client_id from authURL
+		clientID := ""
+		if u2, errP := url.Parse(authURL); errP == nil {
+			clientID = u2.Query().Get("client_id")
+		}
+		// Exchange code for tokens with redirect equal to mgmtRedirect
+		form := url.Values{
+			"grant_type":    {"authorization_code"},
+			"client_id":     {clientID},
+			"code":          {code},
+			"redirect_uri":  {"http://localhost:1455/auth/callback"},
+			"code_verifier": {pkceCodes.CodeVerifier},
+		}
+		httpClient := util.SetProxy(h.cfg, &http.Client{})
+		req, _ := http.NewRequestWithContext(ctx, "POST", "https://auth.openai.com/oauth/token", strings.NewReader(form.Encode()))
+		req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+		req.Header.Set("Accept", "application/json")
+		resp, errDo := httpClient.Do(req)
+		if errDo != nil {
+			authErr := codex.NewAuthenticationError(codex.ErrCodeExchangeFailed, errDo)
+			oauthStatus[state] = "Failed to exchange authorization code for tokens"
+			log.Errorf("Failed to exchange authorization code for tokens: %v", authErr)
+			return
+		}
+		defer func() { _ = resp.Body.Close() }()
+		respBody, _ := io.ReadAll(resp.Body)
+		if resp.StatusCode != http.StatusOK {
+			oauthStatus[state] = fmt.Sprintf("Token exchange failed with status %d", resp.StatusCode)
+			log.Errorf("token exchange failed with status %d: %s", resp.StatusCode, string(respBody))
+			return
+		}
+		var tokenResp struct {
+			AccessToken  string `json:"access_token"`
+			RefreshToken string `json:"refresh_token"`
+			IDToken      string `json:"id_token"`
+			ExpiresIn    int    `json:"expires_in"`
+		}
+		if errU := json.Unmarshal(respBody, &tokenResp); errU != nil {
+			oauthStatus[state] = "Failed to parse token response"
+			log.Errorf("failed to parse token response: %v", errU)
+			return
+		}
+		claims, _ := codex.ParseJWTToken(tokenResp.IDToken)
+		email := ""
+		accountID := ""
+		if claims != nil {
+			email = claims.GetUserEmail()
+			accountID = claims.GetAccountID()
+		}
+		// Build bundle compatible with existing storage
+		bundle := &codex.CodexAuthBundle{
+			TokenData: codex.CodexTokenData{
+				IDToken:      tokenResp.IDToken,
+				AccessToken:  tokenResp.AccessToken,
+				RefreshToken: tokenResp.RefreshToken,
+				AccountID:    accountID,
+				Email:        email,
+				Expire:       time.Now().Add(time.Duration(tokenResp.ExpiresIn) * time.Second).Format(time.RFC3339),
+			},
+			LastRefresh: time.Now().Format(time.RFC3339),
+		}
+
+		// Create token storage and persist
+		tokenStorage := openaiAuth.CreateTokenStorage(bundle)
+		openaiClient, errInit := client.NewCodexClient(h.cfg, tokenStorage)
+		if errInit != nil {
+			oauthStatus[state] = "Failed to initialize Codex client"
+			log.Fatalf("Failed to initialize Codex client: %v", errInit)
+			return
+		}
+		if errSave := openaiClient.SaveTokenToFile(); errSave != nil {
+			oauthStatus[state] = "Failed to save authentication tokens"
+			log.Fatalf("Failed to save authentication tokens: %v", errSave)
+			return
+		}
+		log.Info("Authentication successful!")
+		if bundle.APIKey != "" {
+			log.Info("API key obtained and saved")
+		}
+		log.Info("You can now use Codex services through this CLI")
+		delete(oauthStatus, state)
+	}()
+
+	oauthStatus[state] = ""
+	c.JSON(200, gin.H{"status": "ok", "url": authURL, "state": state})
+}
+
+func (h *Handler) RequestQwenToken(c *gin.Context) {
+	ctx := context.Background()
+
+	log.Info("Initializing Qwen authentication...")
+
+	state := fmt.Sprintf("gem-%d", time.Now().UnixNano())
+	// Initialize Qwen auth service
+	qwenAuth := qwen.NewQwenAuth(h.cfg)
+
+	// Generate authorization URL
+	deviceFlow, err := qwenAuth.InitiateDeviceFlow(ctx)
+	if err != nil {
+		log.Fatalf("Failed to generate authorization URL: %v", err)
+		return
+	}
+	authURL := deviceFlow.VerificationURIComplete
+
+	go func() {
+		log.Info("Waiting for authentication...")
+		tokenData, errPollForToken := qwenAuth.PollForToken(deviceFlow.DeviceCode, deviceFlow.CodeVerifier)
+		if errPollForToken != nil {
+			oauthStatus[state] = "Authentication failed"
+			fmt.Printf("Authentication failed: %v\n", errPollForToken)
+			return
+		}
+
+		// Create token storage
+		tokenStorage := qwenAuth.CreateTokenStorage(tokenData)
+
+		// Initialize Qwen client
+		qwenClient := client.NewQwenClient(h.cfg, tokenStorage)
+
+		tokenStorage.Email = fmt.Sprintf("qwen-%d", time.Now().UnixMilli())
+
+		// Save token storage
+		if err = qwenClient.SaveTokenToFile(); err != nil {
+			log.Fatalf("Failed to save authentication tokens: %v", err)
+			oauthStatus[state] = "Failed to save authentication tokens"
+			return
+		}
+
+		log.Info("Authentication successful!")
+		log.Info("You can now use Qwen services through this CLI")
+		delete(oauthStatus, state)
+	}()
+
+	oauthStatus[state] = ""
+	c.JSON(200, gin.H{"status": "ok", "url": authURL, "state": state})
+}
+
+func (h *Handler) GetAuthStatus(c *gin.Context) {
+	state := c.Query("state")
+	if err, ok := oauthStatus[state]; ok {
+		if err != "" {
+			c.JSON(200, gin.H{"status": "error", "error": err})
+		} else {
+			c.JSON(200, gin.H{"status": "wait"})
+			return
+		}
+	} else {
+		c.JSON(200, gin.H{"status": "ok"})
+	}
+	delete(oauthStatus, state)
+}

internal/api/handlers/management/config_basic.go

@@ -0,0 +1,53 @@
+package management
+
+import (
+	"github.com/gin-gonic/gin"
+)
+
+func (h *Handler) GetConfig(c *gin.Context) {
+	c.JSON(200, h.cfg)
+}
+
+// Debug
+func (h *Handler) GetDebug(c *gin.Context) { c.JSON(200, gin.H{"debug": h.cfg.Debug}) }
+func (h *Handler) PutDebug(c *gin.Context) { h.updateBoolField(c, func(v bool) { h.cfg.Debug = v }) }
+
+// ForceGPT5Codex
+func (h *Handler) GetForceGPT5Codex(c *gin.Context) {
+	c.JSON(200, gin.H{"gpt-5-codex": h.cfg.ForceGPT5Codex})
+}
+func (h *Handler) PutForceGPT5Codex(c *gin.Context) {
+	h.updateBoolField(c, func(v bool) { h.cfg.ForceGPT5Codex = v })
+}
+
+// Request log
+func (h *Handler) GetRequestLog(c *gin.Context) { c.JSON(200, gin.H{"request-log": h.cfg.RequestLog}) }
+func (h *Handler) PutRequestLog(c *gin.Context) {
+	h.updateBoolField(c, func(v bool) { h.cfg.RequestLog = v })
+}
+
+// Request retry
+func (h *Handler) GetRequestRetry(c *gin.Context) {
+	c.JSON(200, gin.H{"request-retry": h.cfg.RequestRetry})
+}
+func (h *Handler) PutRequestRetry(c *gin.Context) {
+	h.updateIntField(c, func(v int) { h.cfg.RequestRetry = v })
+}
+
+// Allow localhost unauthenticated
+func (h *Handler) GetAllowLocalhost(c *gin.Context) {
+	c.JSON(200, gin.H{"allow-localhost-unauthenticated": h.cfg.AllowLocalhostUnauthenticated})
+}
+func (h *Handler) PutAllowLocalhost(c *gin.Context) {
+	h.updateBoolField(c, func(v bool) { h.cfg.AllowLocalhostUnauthenticated = v })
+}
+
+// Proxy URL
+func (h *Handler) GetProxyURL(c *gin.Context) { c.JSON(200, gin.H{"proxy-url": h.cfg.ProxyURL}) }
+func (h *Handler) PutProxyURL(c *gin.Context) {
+	h.updateStringField(c, func(v string) { h.cfg.ProxyURL = v })
+}
+func (h *Handler) DeleteProxyURL(c *gin.Context) {
+	h.cfg.ProxyURL = ""
+	h.persist(c)
+}

internal/api/handlers/management/config_lists.go

@@ -0,0 +1,326 @@
+package management
+
+import (
+	"encoding/json"
+	"fmt"
+
+	"github.com/gin-gonic/gin"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+)
+
+// Generic helpers for list[string]
+func (h *Handler) putStringList(c *gin.Context, set func([]string)) {
+	data, err := c.GetRawData()
+	if err != nil {
+		c.JSON(400, gin.H{"error": "failed to read body"})
+		return
+	}
+	var arr []string
+	if err = json.Unmarshal(data, &arr); err != nil {
+		var obj struct {
+			Items []string `json:"items"`
+		}
+		if err2 := json.Unmarshal(data, &obj); err2 != nil || len(obj.Items) == 0 {
+			c.JSON(400, gin.H{"error": "invalid body"})
+			return
+		}
+		arr = obj.Items
+	}
+	set(arr)
+	h.persist(c)
+}
+
+func (h *Handler) patchStringList(c *gin.Context, target *[]string) {
+	var body struct {
+		Old   *string `json:"old"`
+		New   *string `json:"new"`
+		Index *int    `json:"index"`
+		Value *string `json:"value"`
+	}
+	if err := c.ShouldBindJSON(&body); err != nil {
+		c.JSON(400, gin.H{"error": "invalid body"})
+		return
+	}
+	if body.Index != nil && body.Value != nil && *body.Index >= 0 && *body.Index < len(*target) {
+		(*target)[*body.Index] = *body.Value
+		h.persist(c)
+		return
+	}
+	if body.Old != nil && body.New != nil {
+		for i := range *target {
+			if (*target)[i] == *body.Old {
+				(*target)[i] = *body.New
+				h.persist(c)
+				return
+			}
+		}
+		*target = append(*target, *body.New)
+		h.persist(c)
+		return
+	}
+	c.JSON(400, gin.H{"error": "missing fields"})
+}
+
+func (h *Handler) deleteFromStringList(c *gin.Context, target *[]string) {
+	if idxStr := c.Query("index"); idxStr != "" {
+		var idx int
+		_, err := fmt.Sscanf(idxStr, "%d", &idx)
+		if err == nil && idx >= 0 && idx < len(*target) {
+			*target = append((*target)[:idx], (*target)[idx+1:]...)
+			h.persist(c)
+			return
+		}
+	}
+	if val := c.Query("value"); val != "" {
+		out := make([]string, 0, len(*target))
+		for _, v := range *target {
+			if v != val {
+				out = append(out, v)
+			}
+		}
+		*target = out
+		h.persist(c)
+		return
+	}
+	c.JSON(400, gin.H{"error": "missing index or value"})
+}
+
+// api-keys
+func (h *Handler) GetAPIKeys(c *gin.Context) { c.JSON(200, gin.H{"api-keys": h.cfg.APIKeys}) }
+func (h *Handler) PutAPIKeys(c *gin.Context) {
+	h.putStringList(c, func(v []string) { h.cfg.APIKeys = v })
+}
+func (h *Handler) PatchAPIKeys(c *gin.Context)  { h.patchStringList(c, &h.cfg.APIKeys) }
+func (h *Handler) DeleteAPIKeys(c *gin.Context) { h.deleteFromStringList(c, &h.cfg.APIKeys) }
+
+// generative-language-api-key
+func (h *Handler) GetGlKeys(c *gin.Context) {
+	c.JSON(200, gin.H{"generative-language-api-key": h.cfg.GlAPIKey})
+}
+func (h *Handler) PutGlKeys(c *gin.Context) {
+	h.putStringList(c, func(v []string) { h.cfg.GlAPIKey = v })
+}
+func (h *Handler) PatchGlKeys(c *gin.Context)  { h.patchStringList(c, &h.cfg.GlAPIKey) }
+func (h *Handler) DeleteGlKeys(c *gin.Context) { h.deleteFromStringList(c, &h.cfg.GlAPIKey) }
+
+// claude-api-key: []ClaudeKey
+func (h *Handler) GetClaudeKeys(c *gin.Context) {
+	c.JSON(200, gin.H{"claude-api-key": h.cfg.ClaudeKey})
+}
+func (h *Handler) PutClaudeKeys(c *gin.Context) {
+	data, err := c.GetRawData()
+	if err != nil {
+		c.JSON(400, gin.H{"error": "failed to read body"})
+		return
+	}
+	var arr []config.ClaudeKey
+	if err = json.Unmarshal(data, &arr); err != nil {
+		var obj struct {
+			Items []config.ClaudeKey `json:"items"`
+		}
+		if err2 := json.Unmarshal(data, &obj); err2 != nil || len(obj.Items) == 0 {
+			c.JSON(400, gin.H{"error": "invalid body"})
+			return
+		}
+		arr = obj.Items
+	}
+	h.cfg.ClaudeKey = arr
+	h.persist(c)
+}
+func (h *Handler) PatchClaudeKey(c *gin.Context) {
+	var body struct {
+		Index *int              `json:"index"`
+		Match *string           `json:"match"`
+		Value *config.ClaudeKey `json:"value"`
+	}
+	if err := c.ShouldBindJSON(&body); err != nil || body.Value == nil {
+		c.JSON(400, gin.H{"error": "invalid body"})
+		return
+	}
+	if body.Index != nil && *body.Index >= 0 && *body.Index < len(h.cfg.ClaudeKey) {
+		h.cfg.ClaudeKey[*body.Index] = *body.Value
+		h.persist(c)
+		return
+	}
+	if body.Match != nil {
+		for i := range h.cfg.ClaudeKey {
+			if h.cfg.ClaudeKey[i].APIKey == *body.Match {
+				h.cfg.ClaudeKey[i] = *body.Value
+				h.persist(c)
+				return
+			}
+		}
+	}
+	c.JSON(404, gin.H{"error": "item not found"})
+}
+func (h *Handler) DeleteClaudeKey(c *gin.Context) {
+	if val := c.Query("api-key"); val != "" {
+		out := make([]config.ClaudeKey, 0, len(h.cfg.ClaudeKey))
+		for _, v := range h.cfg.ClaudeKey {
+			if v.APIKey != val {
+				out = append(out, v)
+			}
+		}
+		h.cfg.ClaudeKey = out
+		h.persist(c)
+		return
+	}
+	if idxStr := c.Query("index"); idxStr != "" {
+		var idx int
+		_, err := fmt.Sscanf(idxStr, "%d", &idx)
+		if err == nil && idx >= 0 && idx < len(h.cfg.ClaudeKey) {
+			h.cfg.ClaudeKey = append(h.cfg.ClaudeKey[:idx], h.cfg.ClaudeKey[idx+1:]...)
+			h.persist(c)
+			return
+		}
+	}
+	c.JSON(400, gin.H{"error": "missing api-key or index"})
+}
+
+// openai-compatibility: []OpenAICompatibility
+func (h *Handler) GetOpenAICompat(c *gin.Context) {
+	c.JSON(200, gin.H{"openai-compatibility": h.cfg.OpenAICompatibility})
+}
+func (h *Handler) PutOpenAICompat(c *gin.Context) {
+	data, err := c.GetRawData()
+	if err != nil {
+		c.JSON(400, gin.H{"error": "failed to read body"})
+		return
+	}
+	var arr []config.OpenAICompatibility
+	if err = json.Unmarshal(data, &arr); err != nil {
+		var obj struct {
+			Items []config.OpenAICompatibility `json:"items"`
+		}
+		if err2 := json.Unmarshal(data, &obj); err2 != nil || len(obj.Items) == 0 {
+			c.JSON(400, gin.H{"error": "invalid body"})
+			return
+		}
+		arr = obj.Items
+	}
+	h.cfg.OpenAICompatibility = arr
+	h.persist(c)
+}
+func (h *Handler) PatchOpenAICompat(c *gin.Context) {
+	var body struct {
+		Name  *string                     `json:"name"`
+		Index *int                        `json:"index"`
+		Value *config.OpenAICompatibility `json:"value"`
+	}
+	if err := c.ShouldBindJSON(&body); err != nil || body.Value == nil {
+		c.JSON(400, gin.H{"error": "invalid body"})
+		return
+	}
+	if body.Index != nil && *body.Index >= 0 && *body.Index < len(h.cfg.OpenAICompatibility) {
+		h.cfg.OpenAICompatibility[*body.Index] = *body.Value
+		h.persist(c)
+		return
+	}
+	if body.Name != nil {
+		for i := range h.cfg.OpenAICompatibility {
+			if h.cfg.OpenAICompatibility[i].Name == *body.Name {
+				h.cfg.OpenAICompatibility[i] = *body.Value
+				h.persist(c)
+				return
+			}
+		}
+	}
+	c.JSON(404, gin.H{"error": "item not found"})
+}
+func (h *Handler) DeleteOpenAICompat(c *gin.Context) {
+	if name := c.Query("name"); name != "" {
+		out := make([]config.OpenAICompatibility, 0, len(h.cfg.OpenAICompatibility))
+		for _, v := range h.cfg.OpenAICompatibility {
+			if v.Name != name {
+				out = append(out, v)
+			}
+		}
+		h.cfg.OpenAICompatibility = out
+		h.persist(c)
+		return
+	}
+	if idxStr := c.Query("index"); idxStr != "" {
+		var idx int
+		_, err := fmt.Sscanf(idxStr, "%d", &idx)
+		if err == nil && idx >= 0 && idx < len(h.cfg.OpenAICompatibility) {
+			h.cfg.OpenAICompatibility = append(h.cfg.OpenAICompatibility[:idx], h.cfg.OpenAICompatibility[idx+1:]...)
+			h.persist(c)
+			return
+		}
+	}
+	c.JSON(400, gin.H{"error": "missing name or index"})
+}
+
+// codex-api-key: []CodexKey
+func (h *Handler) GetCodexKeys(c *gin.Context) {
+	c.JSON(200, gin.H{"codex-api-key": h.cfg.CodexKey})
+}
+func (h *Handler) PutCodexKeys(c *gin.Context) {
+	data, err := c.GetRawData()
+	if err != nil {
+		c.JSON(400, gin.H{"error": "failed to read body"})
+		return
+	}
+	var arr []config.CodexKey
+	if err = json.Unmarshal(data, &arr); err != nil {
+		var obj struct {
+			Items []config.CodexKey `json:"items"`
+		}
+		if err2 := json.Unmarshal(data, &obj); err2 != nil || len(obj.Items) == 0 {
+			c.JSON(400, gin.H{"error": "invalid body"})
+			return
+		}
+		arr = obj.Items
+	}
+	h.cfg.CodexKey = arr
+	h.persist(c)
+}
+func (h *Handler) PatchCodexKey(c *gin.Context) {
+	var body struct {
+		Index *int             `json:"index"`
+		Match *string          `json:"match"`
+		Value *config.CodexKey `json:"value"`
+	}
+	if err := c.ShouldBindJSON(&body); err != nil || body.Value == nil {
+		c.JSON(400, gin.H{"error": "invalid body"})
+		return
+	}
+	if body.Index != nil && *body.Index >= 0 && *body.Index < len(h.cfg.CodexKey) {
+		h.cfg.CodexKey[*body.Index] = *body.Value
+		h.persist(c)
+		return
+	}
+	if body.Match != nil {
+		for i := range h.cfg.CodexKey {
+			if h.cfg.CodexKey[i].APIKey == *body.Match {
+				h.cfg.CodexKey[i] = *body.Value
+				h.persist(c)
+				return
+			}
+		}
+	}
+	c.JSON(404, gin.H{"error": "item not found"})
+}
+func (h *Handler) DeleteCodexKey(c *gin.Context) {
+	if val := c.Query("api-key"); val != "" {
+		out := make([]config.CodexKey, 0, len(h.cfg.CodexKey))
+		for _, v := range h.cfg.CodexKey {
+			if v.APIKey != val {
+				out = append(out, v)
+			}
+		}
+		h.cfg.CodexKey = out
+		h.persist(c)
+		return
+	}
+	if idxStr := c.Query("index"); idxStr != "" {
+		var idx int
+		_, err := fmt.Sscanf(idxStr, "%d", &idx)
+		if err == nil && idx >= 0 && idx < len(h.cfg.CodexKey) {
+			h.cfg.CodexKey = append(h.cfg.CodexKey[:idx], h.cfg.CodexKey[idx+1:]...)
+			h.persist(c)
+			return
+		}
+	}
+	c.JSON(400, gin.H{"error": "missing api-key or index"})
+}

internal/api/handlers/management/handler.go

@@ -0,0 +1,199 @@
+// Package management provides the management API handlers and middleware
+// for configuring the server and managing auth files.
+package management
+
+import (
+	"fmt"
+	"net/http"
+	"strings"
+	"sync"
+	"time"
+
+	"github.com/gin-gonic/gin"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	"golang.org/x/crypto/bcrypt"
+)
+
+type attemptInfo struct {
+	count        int
+	blockedUntil time.Time
+}
+
+// Handler aggregates config reference, persistence path and helpers.
+type Handler struct {
+	cfg            *config.Config
+	configFilePath string
+	mu             sync.Mutex
+
+	attemptsMu     sync.Mutex
+	failedAttempts map[string]*attemptInfo // keyed by client IP
+}
+
+// NewHandler creates a new management handler instance.
+func NewHandler(cfg *config.Config, configFilePath string) *Handler {
+	return &Handler{cfg: cfg, configFilePath: configFilePath, failedAttempts: make(map[string]*attemptInfo)}
+}
+
+// SetConfig updates the in-memory config reference when the server hot-reloads.
+func (h *Handler) SetConfig(cfg *config.Config) { h.cfg = cfg }
+
+// Middleware enforces access control for management endpoints.
+// All requests (local and remote) require a valid management key.
+// Additionally, remote access requires allow-remote-management=true.
+func (h *Handler) Middleware() gin.HandlerFunc {
+	const maxFailures = 5
+	const banDuration = 30 * time.Minute
+
+	return func(c *gin.Context) {
+		clientIP := c.ClientIP()
+
+		// For remote IPs, enforce allow-remote-management and ban checks
+		if !(clientIP == "127.0.0.1" || clientIP == "::1") {
+			// Check if IP is currently blocked
+			h.attemptsMu.Lock()
+			ai := h.failedAttempts[clientIP]
+			if ai != nil {
+				if !ai.blockedUntil.IsZero() {
+					if time.Now().Before(ai.blockedUntil) {
+						remaining := time.Until(ai.blockedUntil).Round(time.Second)
+						h.attemptsMu.Unlock()
+						c.AbortWithStatusJSON(http.StatusForbidden, gin.H{"error": fmt.Sprintf("IP banned due to too many failed attempts. Try again in %s", remaining)})
+						return
+					}
+					// Ban expired, reset state
+					ai.blockedUntil = time.Time{}
+					ai.count = 0
+				}
+			}
+			h.attemptsMu.Unlock()
+
+			allowRemote := h.cfg.RemoteManagement.AllowRemote
+			if !allowRemote {
+				allowRemote = true
+			}
+			if !allowRemote {
+				c.AbortWithStatusJSON(http.StatusForbidden, gin.H{"error": "remote management disabled"})
+				return
+			}
+		}
+		secret := h.cfg.RemoteManagement.SecretKey
+		if secret == "" {
+			c.AbortWithStatusJSON(http.StatusForbidden, gin.H{"error": "remote management key not set"})
+			return
+		}
+
+		// Accept either Authorization: Bearer <key> or X-Management-Key
+		var provided string
+		if ah := c.GetHeader("Authorization"); ah != "" {
+			parts := strings.SplitN(ah, " ", 2)
+			if len(parts) == 2 && strings.ToLower(parts[0]) == "bearer" {
+				provided = parts[1]
+			} else {
+				provided = ah
+			}
+		}
+		if provided == "" {
+			provided = c.GetHeader("X-Management-Key")
+		}
+
+		if !(clientIP == "127.0.0.1" || clientIP == "::1") {
+			// For remote IPs, enforce key and track failures
+			fail := func() {
+				h.attemptsMu.Lock()
+				ai := h.failedAttempts[clientIP]
+				if ai == nil {
+					ai = &attemptInfo{}
+					h.failedAttempts[clientIP] = ai
+				}
+				ai.count++
+				if ai.count >= maxFailures {
+					ai.blockedUntil = time.Now().Add(banDuration)
+					ai.count = 0
+				}
+				h.attemptsMu.Unlock()
+			}
+
+			if provided == "" {
+				fail()
+				c.AbortWithStatusJSON(http.StatusUnauthorized, gin.H{"error": "missing management key"})
+				return
+			}
+
+			if err := bcrypt.CompareHashAndPassword([]byte(secret), []byte(provided)); err != nil {
+				fail()
+				c.AbortWithStatusJSON(http.StatusUnauthorized, gin.H{"error": "invalid management key"})
+				return
+			}
+
+			// Success: reset failed count for this IP
+			h.attemptsMu.Lock()
+			if ai := h.failedAttempts[clientIP]; ai != nil {
+				ai.count = 0
+				ai.blockedUntil = time.Time{}
+			}
+			h.attemptsMu.Unlock()
+		}
+
+		c.Next()
+	}
+}
+
+// persist saves the current in-memory config to disk.
+func (h *Handler) persist(c *gin.Context) bool {
+	h.mu.Lock()
+	defer h.mu.Unlock()
+	// Preserve comments when writing
+	if err := config.SaveConfigPreserveComments(h.configFilePath, h.cfg); err != nil {
+		c.JSON(http.StatusInternalServerError, gin.H{"error": fmt.Sprintf("failed to save config: %v", err)})
+		return false
+	}
+	c.JSON(http.StatusOK, gin.H{"status": "ok"})
+	return true
+}
+
+// Helper methods for simple types
+func (h *Handler) updateBoolField(c *gin.Context, set func(bool)) {
+	var body struct {
+		Value *bool `json:"value"`
+	}
+	if err := c.ShouldBindJSON(&body); err != nil || body.Value == nil {
+		var m map[string]any
+		if err2 := c.ShouldBindJSON(&m); err2 == nil {
+			for _, v := range m {
+				if b, ok := v.(bool); ok {
+					set(b)
+					h.persist(c)
+					return
+				}
+			}
+		}
+		c.JSON(http.StatusBadRequest, gin.H{"error": "invalid body"})
+		return
+	}
+	set(*body.Value)
+	h.persist(c)
+}
+
+func (h *Handler) updateIntField(c *gin.Context, set func(int)) {
+	var body struct {
+		Value *int `json:"value"`
+	}
+	if err := c.ShouldBindJSON(&body); err != nil || body.Value == nil {
+		c.JSON(http.StatusBadRequest, gin.H{"error": "invalid body"})
+		return
+	}
+	set(*body.Value)
+	h.persist(c)
+}
+
+func (h *Handler) updateStringField(c *gin.Context, set func(string)) {
+	var body struct {
+		Value *string `json:"value"`
+	}
+	if err := c.ShouldBindJSON(&body); err != nil || body.Value == nil {
+		c.JSON(http.StatusBadRequest, gin.H{"error": "invalid body"})
+		return
+	}
+	set(*body.Value)
+	h.persist(c)
+}

internal/api/handlers/management/quota.go

@@ -0,0 +1,18 @@
+package management
+
+import "github.com/gin-gonic/gin"
+
+// Quota exceeded toggles
+func (h *Handler) GetSwitchProject(c *gin.Context) {
+	c.JSON(200, gin.H{"switch-project": h.cfg.QuotaExceeded.SwitchProject})
+}
+func (h *Handler) PutSwitchProject(c *gin.Context) {
+	h.updateBoolField(c, func(v bool) { h.cfg.QuotaExceeded.SwitchProject = v })
+}
+
+func (h *Handler) GetSwitchPreviewModel(c *gin.Context) {
+	c.JSON(200, gin.H{"switch-preview-model": h.cfg.QuotaExceeded.SwitchPreviewModel})
+}
+func (h *Handler) PutSwitchPreviewModel(c *gin.Context) {
+	h.updateBoolField(c, func(v bool) { h.cfg.QuotaExceeded.SwitchPreviewModel = v })
+}

internal/api/handlers/openai/openai_handlers.go

@@ -8,17 +8,20 @@ package openai
 
 import (
 	"context"
+	"encoding/json"
 	"fmt"
 	"net/http"
 	"time"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/api/handlers"
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/api/handlers"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 	"github.com/tidwall/gjson"
+	"github.com/tidwall/sjson"
 )
 
 // OpenAIAPIHandler contains the handlers for OpenAI API endpoints.
@@ -57,9 +60,33 @@ func (h *OpenAIAPIHandler) Models() []map[string]any {
 // It returns a list of available AI models with their capabilities
 // and specifications in OpenAI-compatible format.
 func (h *OpenAIAPIHandler) OpenAIModels(c *gin.Context) {
+	// Get all available models
+	allModels := h.Models()
+
+	// Filter to only include the 4 required fields: id, object, created, owned_by
+	filteredModels := make([]map[string]any, len(allModels))
+	for i, model := range allModels {
+		filteredModel := map[string]any{
+			"id":     model["id"],
+			"object": model["object"],
+		}
+
+		// Add created field if it exists
+		if created, exists := model["created"]; exists {
+			filteredModel["created"] = created
+		}
+
+		// Add owned_by field if it exists
+		if ownedBy, exists := model["owned_by"]; exists {
+			filteredModel["owned_by"] = ownedBy
+		}
+
+		filteredModels[i] = filteredModel
+	}
+
 	c.JSON(http.StatusOK, gin.H{
 		"object": "list",
-		"data":   h.Models(),
+		"data":   filteredModels,
 	})
 }
 
@@ -92,6 +119,276 @@ func (h *OpenAIAPIHandler) ChatCompletions(c *gin.Context) {
 
 }
 
+// Completions handles the /v1/completions endpoint.
+// It determines whether the request is for a streaming or non-streaming response
+// and calls the appropriate handler based on the model provider.
+// This endpoint follows the OpenAI completions API specification.
+//
+// Parameters:
+//   - c: The Gin context containing the HTTP request and response
+func (h *OpenAIAPIHandler) Completions(c *gin.Context) {
+	rawJSON, err := c.GetRawData()
+	// If data retrieval fails, return a 400 Bad Request error.
+	if err != nil {
+		c.JSON(http.StatusBadRequest, handlers.ErrorResponse{
+			Error: handlers.ErrorDetail{
+				Message: fmt.Sprintf("Invalid request: %v", err),
+				Type:    "invalid_request_error",
+			},
+		})
+		return
+	}
+
+	// Check if the client requested a streaming response.
+	streamResult := gjson.GetBytes(rawJSON, "stream")
+	if streamResult.Type == gjson.True {
+		h.handleCompletionsStreamingResponse(c, rawJSON)
+	} else {
+		h.handleCompletionsNonStreamingResponse(c, rawJSON)
+	}
+
+}
+
+// convertCompletionsRequestToChatCompletions converts OpenAI completions API request to chat completions format.
+// This allows the completions endpoint to use the existing chat completions infrastructure.
+//
+// Parameters:
+//   - rawJSON: The raw JSON bytes of the completions request
+//
+// Returns:
+//   - []byte: The converted chat completions request
+func convertCompletionsRequestToChatCompletions(rawJSON []byte) []byte {
+	root := gjson.ParseBytes(rawJSON)
+
+	// Extract prompt from completions request
+	prompt := root.Get("prompt").String()
+	if prompt == "" {
+		prompt = "Complete this:"
+	}
+
+	// Create chat completions structure
+	out := `{"model":"","messages":[{"role":"user","content":""}]}`
+
+	// Set model
+	if model := root.Get("model"); model.Exists() {
+		out, _ = sjson.Set(out, "model", model.String())
+	}
+
+	// Set the prompt as user message content
+	out, _ = sjson.Set(out, "messages.0.content", prompt)
+
+	// Copy other parameters from completions to chat completions
+	if maxTokens := root.Get("max_tokens"); maxTokens.Exists() {
+		out, _ = sjson.Set(out, "max_tokens", maxTokens.Int())
+	}
+
+	if temperature := root.Get("temperature"); temperature.Exists() {
+		out, _ = sjson.Set(out, "temperature", temperature.Float())
+	}
+
+	if topP := root.Get("top_p"); topP.Exists() {
+		out, _ = sjson.Set(out, "top_p", topP.Float())
+	}
+
+	if frequencyPenalty := root.Get("frequency_penalty"); frequencyPenalty.Exists() {
+		out, _ = sjson.Set(out, "frequency_penalty", frequencyPenalty.Float())
+	}
+
+	if presencePenalty := root.Get("presence_penalty"); presencePenalty.Exists() {
+		out, _ = sjson.Set(out, "presence_penalty", presencePenalty.Float())
+	}
+
+	if stop := root.Get("stop"); stop.Exists() {
+		out, _ = sjson.SetRaw(out, "stop", stop.Raw)
+	}
+
+	if stream := root.Get("stream"); stream.Exists() {
+		out, _ = sjson.Set(out, "stream", stream.Bool())
+	}
+
+	if logprobs := root.Get("logprobs"); logprobs.Exists() {
+		out, _ = sjson.Set(out, "logprobs", logprobs.Bool())
+	}
+
+	if topLogprobs := root.Get("top_logprobs"); topLogprobs.Exists() {
+		out, _ = sjson.Set(out, "top_logprobs", topLogprobs.Int())
+	}
+
+	if echo := root.Get("echo"); echo.Exists() {
+		out, _ = sjson.Set(out, "echo", echo.Bool())
+	}
+
+	return []byte(out)
+}
+
+// convertChatCompletionsResponseToCompletions converts chat completions API response back to completions format.
+// This ensures the completions endpoint returns data in the expected format.
+//
+// Parameters:
+//   - rawJSON: The raw JSON bytes of the chat completions response
+//
+// Returns:
+//   - []byte: The converted completions response
+func convertChatCompletionsResponseToCompletions(rawJSON []byte) []byte {
+	root := gjson.ParseBytes(rawJSON)
+
+	// Base completions response structure
+	out := `{"id":"","object":"text_completion","created":0,"model":"","choices":[]}`
+
+	// Copy basic fields
+	if id := root.Get("id"); id.Exists() {
+		out, _ = sjson.Set(out, "id", id.String())
+	}
+
+	if created := root.Get("created"); created.Exists() {
+		out, _ = sjson.Set(out, "created", created.Int())
+	}
+
+	if model := root.Get("model"); model.Exists() {
+		out, _ = sjson.Set(out, "model", model.String())
+	}
+
+	if usage := root.Get("usage"); usage.Exists() {
+		out, _ = sjson.SetRaw(out, "usage", usage.Raw)
+	}
+
+	// Convert choices from chat completions to completions format
+	var choices []interface{}
+	if chatChoices := root.Get("choices"); chatChoices.Exists() && chatChoices.IsArray() {
+		chatChoices.ForEach(func(_, choice gjson.Result) bool {
+			completionsChoice := map[string]interface{}{
+				"index": choice.Get("index").Int(),
+			}
+
+			// Extract text content from message.content
+			if message := choice.Get("message"); message.Exists() {
+				if content := message.Get("content"); content.Exists() {
+					completionsChoice["text"] = content.String()
+				}
+			} else if delta := choice.Get("delta"); delta.Exists() {
+				// For streaming responses, use delta.content
+				if content := delta.Get("content"); content.Exists() {
+					completionsChoice["text"] = content.String()
+				}
+			}
+
+			// Copy finish_reason
+			if finishReason := choice.Get("finish_reason"); finishReason.Exists() {
+				completionsChoice["finish_reason"] = finishReason.String()
+			}
+
+			// Copy logprobs if present
+			if logprobs := choice.Get("logprobs"); logprobs.Exists() {
+				completionsChoice["logprobs"] = logprobs.Value()
+			}
+
+			choices = append(choices, completionsChoice)
+			return true
+		})
+	}
+
+	if len(choices) > 0 {
+		choicesJSON, _ := json.Marshal(choices)
+		out, _ = sjson.SetRaw(out, "choices", string(choicesJSON))
+	}
+
+	return []byte(out)
+}
+
+// convertChatCompletionsStreamChunkToCompletions converts a streaming chat completions chunk to completions format.
+// This handles the real-time conversion of streaming response chunks and filters out empty text responses.
+//
+// Parameters:
+//   - chunkData: The raw JSON bytes of a single chat completions stream chunk
+//
+// Returns:
+//   - []byte: The converted completions stream chunk, or nil if should be filtered out
+func convertChatCompletionsStreamChunkToCompletions(chunkData []byte) []byte {
+	root := gjson.ParseBytes(chunkData)
+
+	// Check if this chunk has any meaningful content
+	hasContent := false
+	if chatChoices := root.Get("choices"); chatChoices.Exists() && chatChoices.IsArray() {
+		chatChoices.ForEach(func(_, choice gjson.Result) bool {
+			// Check if delta has content or finish_reason
+			if delta := choice.Get("delta"); delta.Exists() {
+				if content := delta.Get("content"); content.Exists() && content.String() != "" {
+					hasContent = true
+					return false // Break out of forEach
+				}
+			}
+			// Also check for finish_reason to ensure we don't skip final chunks
+			if finishReason := choice.Get("finish_reason"); finishReason.Exists() && finishReason.String() != "" && finishReason.String() != "null" {
+				hasContent = true
+				return false // Break out of forEach
+			}
+			return true
+		})
+	}
+
+	// If no meaningful content, return nil to indicate this chunk should be skipped
+	if !hasContent {
+		return nil
+	}
+
+	// Base completions stream response structure
+	out := `{"id":"","object":"text_completion","created":0,"model":"","choices":[]}`
+
+	// Copy basic fields
+	if id := root.Get("id"); id.Exists() {
+		out, _ = sjson.Set(out, "id", id.String())
+	}
+
+	if created := root.Get("created"); created.Exists() {
+		out, _ = sjson.Set(out, "created", created.Int())
+	}
+
+	if model := root.Get("model"); model.Exists() {
+		out, _ = sjson.Set(out, "model", model.String())
+	}
+
+	// Convert choices from chat completions delta to completions format
+	var choices []interface{}
+	if chatChoices := root.Get("choices"); chatChoices.Exists() && chatChoices.IsArray() {
+		chatChoices.ForEach(func(_, choice gjson.Result) bool {
+			completionsChoice := map[string]interface{}{
+				"index": choice.Get("index").Int(),
+			}
+
+			// Extract text content from delta.content
+			if delta := choice.Get("delta"); delta.Exists() {
+				if content := delta.Get("content"); content.Exists() && content.String() != "" {
+					completionsChoice["text"] = content.String()
+				} else {
+					completionsChoice["text"] = ""
+				}
+			} else {
+				completionsChoice["text"] = ""
+			}
+
+			// Copy finish_reason
+			if finishReason := choice.Get("finish_reason"); finishReason.Exists() && finishReason.String() != "null" {
+				completionsChoice["finish_reason"] = finishReason.String()
+			}
+
+			// Copy logprobs if present
+			if logprobs := choice.Get("logprobs"); logprobs.Exists() {
+				completionsChoice["logprobs"] = logprobs.Value()
+			}
+
+			choices = append(choices, completionsChoice)
+			return true
+		})
+	}
+
+	if len(choices) > 0 {
+		choicesJSON, _ := json.Marshal(choices)
+		out, _ = sjson.SetRaw(out, "choices", string(choicesJSON))
+	}
+
+	return []byte(out)
+}
+
 // handleNonStreamingResponse handles non-streaming chat completion responses
 // for Gemini models. It selects a client from the pool, sends the request, and
 // aggregates the response before sending it back to the client in OpenAI format.
@@ -108,13 +405,15 @@ func (h *OpenAIAPIHandler) handleNonStreamingResponse(c *gin.Context, rawJSON []
 	var cliClient interfaces.Client
 	defer func() {
 		if cliClient != nil {
-			cliClient.GetRequestMutex().Unlock()
+			if mutex := cliClient.GetRequestMutex(); mutex != nil {
+				mutex.Unlock()
+			}
 		}
 	}()
 
+	var errorResponse *interfaces.ErrorMessage
 	retryCount := 0
 	for retryCount <= h.Cfg.RequestRetry {
-		var errorResponse *interfaces.ErrorMessage
 		cliClient, errorResponse = h.GetClient(modelName)
 		if errorResponse != nil {
 			c.Status(errorResponse.StatusCode)
@@ -125,6 +424,9 @@ func (h *OpenAIAPIHandler) handleNonStreamingResponse(c *gin.Context, rawJSON []
 
 		resp, err := cliClient.SendRawMessage(cliCtx, modelName, rawJSON, "")
 		if err != nil {
+			errorResponse = err
+			h.LoggingAPIResponseError(cliCtx, err)
+
 			switch err.StatusCode {
 			case 429:
 				if h.Cfg.QuotaExceeded.SwitchProject {
@@ -135,6 +437,18 @@ func (h *OpenAIAPIHandler) handleNonStreamingResponse(c *gin.Context, rawJSON []
 				log.Debugf("http status code %d, switch client", err.StatusCode)
 				retryCount++
 				continue
+			case 401:
+				log.Debugf("unauthorized request, try to refresh token, %s", util.HideAPIKey(cliClient.GetEmail()))
+				errRefreshTokens := cliClient.RefreshTokens(cliCtx)
+				if errRefreshTokens != nil {
+					log.Debugf("refresh token failed, switch client, %s", util.HideAPIKey(cliClient.GetEmail()))
+					cliClient.SetUnavailable()
+				}
+				retryCount++
+				continue
+			case 402:
+				cliClient.SetUnavailable()
+				continue
 			default:
 				// Forward other errors directly to the client
 				c.Status(err.StatusCode)
@@ -144,10 +458,16 @@ func (h *OpenAIAPIHandler) handleNonStreamingResponse(c *gin.Context, rawJSON []
 			break
 		} else {
 			_, _ = c.Writer.Write(resp)
-			cliCancel(resp)
+			cliCancel()
 			break
 		}
 	}
+	if errorResponse != nil {
+		c.Status(errorResponse.StatusCode)
+		_, _ = c.Writer.Write([]byte(errorResponse.Error.Error()))
+		cliCancel(errorResponse.Error)
+		return
+	}
 }
 
 // handleStreamingResponse handles streaming responses for Gemini models.
@@ -182,14 +502,16 @@ func (h *OpenAIAPIHandler) handleStreamingResponse(c *gin.Context, rawJSON []byt
 	defer func() {
 		// Ensure the client's mutex is unlocked on function exit.
 		if cliClient != nil {
-			cliClient.GetRequestMutex().Unlock()
+			if mutex := cliClient.GetRequestMutex(); mutex != nil {
+				mutex.Unlock()
+			}
 		}
 	}()
 
+	var errorResponse *interfaces.ErrorMessage
 	retryCount := 0
 outLoop:
 	for retryCount <= h.Cfg.RequestRetry {
-		var errorResponse *interfaces.ErrorMessage
 		cliClient, errorResponse = h.GetClient(modelName)
 		if errorResponse != nil {
 			c.Status(errorResponse.StatusCode)
@@ -207,7 +529,7 @@ outLoop:
 			// Handle client disconnection.
 			case <-c.Request.Context().Done():
 				if c.Request.Context().Err().Error() == "context canceled" {
-					log.Debugf("qwen client disconnected: %v", c.Request.Context().Err())
+					log.Debugf("openai client disconnected: %v", c.Request.Context().Err())
 					cliCancel() // Cancel the backend request.
 					return
 				}
@@ -226,6 +548,9 @@ outLoop:
 			// Handle errors from the backend.
 			case err, okError := <-errChan:
 				if okError {
+					errorResponse = err
+					h.LoggingAPIResponseError(cliCtx, err)
+
 					switch err.StatusCode {
 					case 429:
 						if h.Cfg.QuotaExceeded.SwitchProject {
@@ -236,6 +561,18 @@ outLoop:
 						log.Debugf("http status code %d, switch client", err.StatusCode)
 						retryCount++
 						continue outLoop
+					case 401:
+						log.Debugf("unauthorized request, try to refresh token, %s", util.HideAPIKey(cliClient.GetEmail()))
+						errRefreshTokens := cliClient.RefreshTokens(cliCtx)
+						if errRefreshTokens != nil {
+							log.Debugf("refresh token failed, switch client, %s", util.HideAPIKey(cliClient.GetEmail()))
+							cliClient.SetUnavailable()
+						}
+						retryCount++
+						continue outLoop
+					case 402:
+						cliClient.SetUnavailable()
+						continue outLoop
 					default:
 						// Forward other errors directly to the client
 						c.Status(err.StatusCode)
@@ -250,4 +587,233 @@ outLoop:
 			}
 		}
 	}
+	if errorResponse != nil {
+		c.Status(errorResponse.StatusCode)
+		_, _ = fmt.Fprint(c.Writer, errorResponse.Error.Error())
+		flusher.Flush()
+		cliCancel(errorResponse.Error)
+		return
+	}
+}
+
+// handleCompletionsNonStreamingResponse handles non-streaming completions responses.
+// It converts completions request to chat completions format, sends to backend,
+// then converts the response back to completions format before sending to client.
+//
+// Parameters:
+//   - c: The Gin context containing the HTTP request and response
+//   - rawJSON: The raw JSON bytes of the OpenAI-compatible completions request
+func (h *OpenAIAPIHandler) handleCompletionsNonStreamingResponse(c *gin.Context, rawJSON []byte) {
+	c.Header("Content-Type", "application/json")
+
+	// Convert completions request to chat completions format
+	chatCompletionsJSON := convertCompletionsRequestToChatCompletions(rawJSON)
+
+	modelName := gjson.GetBytes(chatCompletionsJSON, "model").String()
+	cliCtx, cliCancel := h.GetContextWithCancel(h, c, context.Background())
+
+	var cliClient interfaces.Client
+	defer func() {
+		if cliClient != nil {
+			if mutex := cliClient.GetRequestMutex(); mutex != nil {
+				mutex.Unlock()
+			}
+		}
+	}()
+
+	var errorResponse *interfaces.ErrorMessage
+	retryCount := 0
+	for retryCount <= h.Cfg.RequestRetry {
+		cliClient, errorResponse = h.GetClient(modelName)
+		if errorResponse != nil {
+			c.Status(errorResponse.StatusCode)
+			_, _ = fmt.Fprint(c.Writer, errorResponse.Error.Error())
+			cliCancel()
+			return
+		}
+
+		// Send the converted chat completions request
+		resp, err := cliClient.SendRawMessage(cliCtx, modelName, chatCompletionsJSON, "")
+		if err != nil {
+			errorResponse = err
+			h.LoggingAPIResponseError(cliCtx, err)
+
+			switch err.StatusCode {
+			case 429:
+				if h.Cfg.QuotaExceeded.SwitchProject {
+					log.Debugf("quota exceeded, switch client")
+					continue // Restart the client selection process
+				}
+			case 403, 408, 500, 502, 503, 504:
+				log.Debugf("http status code %d, switch client", err.StatusCode)
+				retryCount++
+				continue
+			case 401:
+				log.Debugf("unauthorized request, try to refresh token, %s", util.HideAPIKey(cliClient.GetEmail()))
+				errRefreshTokens := cliClient.RefreshTokens(cliCtx)
+				if errRefreshTokens != nil {
+					log.Debugf("refresh token failed, switch client, %s", util.HideAPIKey(cliClient.GetEmail()))
+					cliClient.SetUnavailable()
+				}
+				retryCount++
+				continue
+			case 402:
+				cliClient.SetUnavailable()
+				continue
+			default:
+				// Forward other errors directly to the client
+				c.Status(err.StatusCode)
+				_, _ = c.Writer.Write([]byte(err.Error.Error()))
+				cliCancel(err.Error)
+			}
+			break
+		} else {
+			// Convert chat completions response back to completions format
+			completionsResp := convertChatCompletionsResponseToCompletions(resp)
+			_, _ = c.Writer.Write(completionsResp)
+			cliCancel()
+			break
+		}
+	}
+	if errorResponse != nil {
+		c.Status(errorResponse.StatusCode)
+		_, _ = c.Writer.Write([]byte(errorResponse.Error.Error()))
+		cliCancel(errorResponse.Error)
+		return
+	}
+
+}
+
+// handleCompletionsStreamingResponse handles streaming completions responses.
+// It converts completions request to chat completions format, streams from backend,
+// then converts each response chunk back to completions format before sending to client.
+//
+// Parameters:
+//   - c: The Gin context containing the HTTP request and response
+//   - rawJSON: The raw JSON bytes of the OpenAI-compatible completions request
+func (h *OpenAIAPIHandler) handleCompletionsStreamingResponse(c *gin.Context, rawJSON []byte) {
+	c.Header("Content-Type", "text/event-stream")
+	c.Header("Cache-Control", "no-cache")
+	c.Header("Connection", "keep-alive")
+	c.Header("Access-Control-Allow-Origin", "*")
+
+	// Get the http.Flusher interface to manually flush the response.
+	flusher, ok := c.Writer.(http.Flusher)
+	if !ok {
+		c.JSON(http.StatusInternalServerError, handlers.ErrorResponse{
+			Error: handlers.ErrorDetail{
+				Message: "Streaming not supported",
+				Type:    "server_error",
+			},
+		})
+		return
+	}
+
+	// Convert completions request to chat completions format
+	chatCompletionsJSON := convertCompletionsRequestToChatCompletions(rawJSON)
+
+	modelName := gjson.GetBytes(chatCompletionsJSON, "model").String()
+	cliCtx, cliCancel := h.GetContextWithCancel(h, c, context.Background())
+
+	var cliClient interfaces.Client
+	defer func() {
+		// Ensure the client's mutex is unlocked on function exit.
+		if cliClient != nil {
+			if mutex := cliClient.GetRequestMutex(); mutex != nil {
+				mutex.Unlock()
+			}
+		}
+	}()
+
+	var errorResponse *interfaces.ErrorMessage
+	retryCount := 0
+outLoop:
+	for retryCount <= h.Cfg.RequestRetry {
+		cliClient, errorResponse = h.GetClient(modelName)
+		if errorResponse != nil {
+			c.Status(errorResponse.StatusCode)
+			_, _ = fmt.Fprint(c.Writer, errorResponse.Error.Error())
+			flusher.Flush()
+			cliCancel()
+			return
+		}
+
+		// Send the converted chat completions request and receive response chunks
+		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, modelName, chatCompletionsJSON, "")
+
+		for {
+			select {
+			// Handle client disconnection.
+			case <-c.Request.Context().Done():
+				if c.Request.Context().Err().Error() == "context canceled" {
+					log.Debugf("client disconnected: %v", c.Request.Context().Err())
+					cliCancel() // Cancel the backend request.
+					return
+				}
+			// Process incoming response chunks.
+			case chunk, okStream := <-respChan:
+				if !okStream {
+					// Stream is closed, send the final [DONE] message.
+					_, _ = fmt.Fprintf(c.Writer, "data: [DONE]\n\n")
+					flusher.Flush()
+					cliCancel()
+					return
+				}
+
+				// Convert chat completions chunk to completions chunk format
+				completionsChunk := convertChatCompletionsStreamChunkToCompletions(chunk)
+				// Skip this chunk if it has no meaningful content (empty text)
+				if completionsChunk != nil {
+					_, _ = fmt.Fprintf(c.Writer, "data: %s\n\n", string(completionsChunk))
+					flusher.Flush()
+				}
+			// Handle errors from the backend.
+			case err, okError := <-errChan:
+				if okError {
+					errorResponse = err
+					h.LoggingAPIResponseError(cliCtx, err)
+
+					switch err.StatusCode {
+					case 429:
+						if h.Cfg.QuotaExceeded.SwitchProject {
+							log.Debugf("quota exceeded, switch client")
+							continue outLoop // Restart the client selection process
+						}
+					case 403, 408, 500, 502, 503, 504:
+						log.Debugf("http status code %d, switch client", err.StatusCode)
+						retryCount++
+						continue outLoop
+					case 401:
+						log.Debugf("unauthorized request, try to refresh token, %s", util.HideAPIKey(cliClient.GetEmail()))
+						errRefreshTokens := cliClient.RefreshTokens(cliCtx)
+						if errRefreshTokens != nil {
+							log.Debugf("refresh token failed, switch client, %s", util.HideAPIKey(cliClient.GetEmail()))
+							cliClient.SetUnavailable()
+						}
+						retryCount++
+						continue outLoop
+					case 402:
+						cliClient.SetUnavailable()
+						continue outLoop
+					default:
+						// Forward other errors directly to the client
+						c.Status(err.StatusCode)
+						_, _ = fmt.Fprint(c.Writer, err.Error.Error())
+						flusher.Flush()
+						cliCancel(err.Error)
+					}
+					return
+				}
+			// Send a keep-alive signal to the client.
+			case <-time.After(500 * time.Millisecond):
+			}
+		}
+	}
+	if errorResponse != nil {
+		c.Status(errorResponse.StatusCode)
+		_, _ = fmt.Fprint(c.Writer, errorResponse.Error.Error())
+		flusher.Flush()
+		cliCancel(errorResponse.Error)
+		return
+	}
 }

internal/api/handlers/openai/openai_responses_handlers.go

@@ -0,0 +1,301 @@
+// Package openai provides HTTP handlers for OpenAIResponses API endpoints.
+// This package implements the OpenAIResponses-compatible API interface, including model listing
+// and chat completion functionality. It supports both streaming and non-streaming responses,
+// and manages a pool of clients to interact with backend services.
+// The handlers translate OpenAIResponses API requests to the appropriate backend format and
+// convert responses back to OpenAIResponses-compatible format.
+package openai
+
+import (
+	"context"
+	"fmt"
+	"net/http"
+	"time"
+
+	"github.com/gin-gonic/gin"
+	"github.com/luispater/CLIProxyAPI/v5/internal/api/handlers"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
+	log "github.com/sirupsen/logrus"
+	"github.com/tidwall/gjson"
+)
+
+// OpenAIResponsesAPIHandler contains the handlers for OpenAIResponses API endpoints.
+// It holds a pool of clients to interact with the backend service.
+type OpenAIResponsesAPIHandler struct {
+	*handlers.BaseAPIHandler
+}
+
+// NewOpenAIResponsesAPIHandler creates a new OpenAIResponses API handlers instance.
+// It takes an BaseAPIHandler instance as input and returns an OpenAIResponsesAPIHandler.
+//
+// Parameters:
+//   - apiHandlers: The base API handlers instance
+//
+// Returns:
+//   - *OpenAIResponsesAPIHandler: A new OpenAIResponses API handlers instance
+func NewOpenAIResponsesAPIHandler(apiHandlers *handlers.BaseAPIHandler) *OpenAIResponsesAPIHandler {
+	return &OpenAIResponsesAPIHandler{
+		BaseAPIHandler: apiHandlers,
+	}
+}
+
+// HandlerType returns the identifier for this handler implementation.
+func (h *OpenAIResponsesAPIHandler) HandlerType() string {
+	return OPENAI_RESPONSE
+}
+
+// Models returns the OpenAIResponses-compatible model metadata supported by this handler.
+func (h *OpenAIResponsesAPIHandler) Models() []map[string]any {
+	// Get dynamic models from the global registry
+	modelRegistry := registry.GetGlobalRegistry()
+	return modelRegistry.GetAvailableModels("openai")
+}
+
+// OpenAIResponsesModels handles the /v1/models endpoint.
+// It returns a list of available AI models with their capabilities
+// and specifications in OpenAIResponses-compatible format.
+func (h *OpenAIResponsesAPIHandler) OpenAIResponsesModels(c *gin.Context) {
+	c.JSON(http.StatusOK, gin.H{
+		"object": "list",
+		"data":   h.Models(),
+	})
+}
+
+// Responses handles the /v1/responses endpoint.
+// It determines whether the request is for a streaming or non-streaming response
+// and calls the appropriate handler based on the model provider.
+//
+// Parameters:
+//   - c: The Gin context containing the HTTP request and response
+func (h *OpenAIResponsesAPIHandler) Responses(c *gin.Context) {
+	rawJSON, err := c.GetRawData()
+	// If data retrieval fails, return a 400 Bad Request error.
+	if err != nil {
+		c.JSON(http.StatusBadRequest, handlers.ErrorResponse{
+			Error: handlers.ErrorDetail{
+				Message: fmt.Sprintf("Invalid request: %v", err),
+				Type:    "invalid_request_error",
+			},
+		})
+		return
+	}
+
+	// Check if the client requested a streaming response.
+	streamResult := gjson.GetBytes(rawJSON, "stream")
+	if streamResult.Type == gjson.True {
+		h.handleStreamingResponse(c, rawJSON)
+	} else {
+		h.handleNonStreamingResponse(c, rawJSON)
+	}
+
+}
+
+// handleNonStreamingResponse handles non-streaming chat completion responses
+// for Gemini models. It selects a client from the pool, sends the request, and
+// aggregates the response before sending it back to the client in OpenAIResponses format.
+//
+// Parameters:
+//   - c: The Gin context containing the HTTP request and response
+//   - rawJSON: The raw JSON bytes of the OpenAIResponses-compatible request
+func (h *OpenAIResponsesAPIHandler) handleNonStreamingResponse(c *gin.Context, rawJSON []byte) {
+	c.Header("Content-Type", "application/json")
+
+	modelName := gjson.GetBytes(rawJSON, "model").String()
+	cliCtx, cliCancel := h.GetContextWithCancel(h, c, context.Background())
+
+	var cliClient interfaces.Client
+	defer func() {
+		if cliClient != nil {
+			if mutex := cliClient.GetRequestMutex(); mutex != nil {
+				mutex.Unlock()
+			}
+		}
+	}()
+
+	var errorResponse *interfaces.ErrorMessage
+	retryCount := 0
+	for retryCount <= h.Cfg.RequestRetry {
+		cliClient, errorResponse = h.GetClient(modelName)
+		if errorResponse != nil {
+			c.Status(errorResponse.StatusCode)
+			_, _ = fmt.Fprint(c.Writer, errorResponse.Error.Error())
+			cliCancel()
+			return
+		}
+
+		resp, err := cliClient.SendRawMessage(cliCtx, modelName, rawJSON, "")
+		if err != nil {
+			errorResponse = err
+			h.LoggingAPIResponseError(cliCtx, err)
+
+			switch err.StatusCode {
+			case 429:
+				if h.Cfg.QuotaExceeded.SwitchProject {
+					log.Debugf("quota exceeded, switch client")
+					continue // Restart the client selection process
+				}
+			case 403, 408, 500, 502, 503, 504:
+				log.Debugf("http status code %d, switch client", err.StatusCode)
+				retryCount++
+				continue
+			case 401:
+				log.Debugf("unauthorized request, try to refresh token, %s", util.HideAPIKey(cliClient.GetEmail()))
+				errRefreshTokens := cliClient.RefreshTokens(cliCtx)
+				if errRefreshTokens != nil {
+					log.Debugf("refresh token failed, switch client, %s", util.HideAPIKey(cliClient.GetEmail()))
+					cliClient.SetUnavailable()
+				}
+				retryCount++
+				continue
+			case 402:
+				cliClient.SetUnavailable()
+				continue
+			default:
+				// Forward other errors directly to the client
+				c.Status(err.StatusCode)
+				_, _ = c.Writer.Write([]byte(err.Error.Error()))
+				cliCancel(err.Error)
+			}
+			break
+		} else {
+			_, _ = c.Writer.Write(resp)
+			cliCancel()
+			break
+		}
+	}
+	if errorResponse != nil {
+		c.Status(errorResponse.StatusCode)
+		_, _ = c.Writer.Write([]byte(errorResponse.Error.Error()))
+		cliCancel(errorResponse.Error)
+		return
+	}
+
+}
+
+// handleStreamingResponse handles streaming responses for Gemini models.
+// It establishes a streaming connection with the backend service and forwards
+// the response chunks to the client in real-time using Server-Sent Events.
+//
+// Parameters:
+//   - c: The Gin context containing the HTTP request and response
+//   - rawJSON: The raw JSON bytes of the OpenAIResponses-compatible request
+func (h *OpenAIResponsesAPIHandler) handleStreamingResponse(c *gin.Context, rawJSON []byte) {
+	c.Header("Content-Type", "text/event-stream")
+	c.Header("Cache-Control", "no-cache")
+	c.Header("Connection", "keep-alive")
+	c.Header("Access-Control-Allow-Origin", "*")
+
+	// Get the http.Flusher interface to manually flush the response.
+	flusher, ok := c.Writer.(http.Flusher)
+	if !ok {
+		c.JSON(http.StatusInternalServerError, handlers.ErrorResponse{
+			Error: handlers.ErrorDetail{
+				Message: "Streaming not supported",
+				Type:    "server_error",
+			},
+		})
+		return
+	}
+
+	modelName := gjson.GetBytes(rawJSON, "model").String()
+	cliCtx, cliCancel := h.GetContextWithCancel(h, c, context.Background())
+
+	var cliClient interfaces.Client
+	defer func() {
+		// Ensure the client's mutex is unlocked on function exit.
+		if cliClient != nil {
+			if mutex := cliClient.GetRequestMutex(); mutex != nil {
+				mutex.Unlock()
+			}
+		}
+	}()
+
+	var errorResponse *interfaces.ErrorMessage
+	retryCount := 0
+outLoop:
+	for retryCount <= h.Cfg.RequestRetry {
+		cliClient, errorResponse = h.GetClient(modelName)
+		if errorResponse != nil {
+			c.Status(errorResponse.StatusCode)
+			_, _ = fmt.Fprint(c.Writer, errorResponse.Error.Error())
+			flusher.Flush()
+			cliCancel()
+			return
+		}
+
+		// Send the message and receive response chunks and errors via channels.
+		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, modelName, rawJSON, "")
+
+		for {
+			select {
+			// Handle client disconnection.
+			case <-c.Request.Context().Done():
+				if c.Request.Context().Err().Error() == "context canceled" {
+					log.Debugf("openai client disconnected: %v", c.Request.Context().Err())
+					cliCancel() // Cancel the backend request.
+					return
+				}
+			// Process incoming response chunks.
+			case chunk, okStream := <-respChan:
+				if !okStream {
+					flusher.Flush()
+					cliCancel()
+					return
+				}
+
+				_, _ = c.Writer.Write(chunk)
+				_, _ = c.Writer.Write([]byte("\n"))
+				flusher.Flush()
+			// Handle errors from the backend.
+			case err, okError := <-errChan:
+				if okError {
+					errorResponse = err
+					h.LoggingAPIResponseError(cliCtx, err)
+					switch err.StatusCode {
+					case 429:
+						if h.Cfg.QuotaExceeded.SwitchProject {
+							log.Debugf("quota exceeded, switch client")
+							continue outLoop // Restart the client selection process
+						}
+					case 403, 408, 500, 502, 503, 504:
+						log.Debugf("http status code %d, switch client", err.StatusCode)
+						retryCount++
+						continue outLoop
+					case 401:
+						log.Debugf("unauthorized request, try to refresh token, %s", util.HideAPIKey(cliClient.GetEmail()))
+						errRefreshTokens := cliClient.RefreshTokens(cliCtx)
+						if errRefreshTokens != nil {
+							log.Debugf("refresh token failed, switch client, %s", util.HideAPIKey(cliClient.GetEmail()))
+							cliClient.SetUnavailable()
+						}
+						retryCount++
+						continue outLoop
+					case 402:
+						cliClient.SetUnavailable()
+						continue outLoop
+					default:
+						// Forward other errors directly to the client
+						c.Status(err.StatusCode)
+						_, _ = fmt.Fprint(c.Writer, err.Error.Error())
+						flusher.Flush()
+						cliCancel(err.Error)
+					}
+					return
+				}
+			// Send a keep-alive signal to the client.
+			case <-time.After(500 * time.Millisecond):
+			}
+		}
+	}
+
+	if errorResponse != nil {
+		c.Status(errorResponse.StatusCode)
+		_, _ = fmt.Fprint(c.Writer, errorResponse.Error.Error())
+		flusher.Flush()
+		cliCancel(errorResponse.Error)
+		return
+	}
+}

internal/api/middleware/request_logging.go

@@ -8,7 +8,7 @@ import (
 	"io"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/logging"
+	"github.com/luispater/CLIProxyAPI/v5/internal/logging"
 )
 
 // RequestLoggingMiddleware creates a Gin middleware that logs HTTP requests and responses.

internal/api/middleware/response_writer.go

@@ -8,7 +8,8 @@ import (
 	"strings"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/logging"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/logging"
 )
 
 // RequestInfo holds essential details of an incoming HTTP request for logging purposes.
@@ -27,6 +28,7 @@ type ResponseWriterWrapper struct {
 	isStreaming  bool                       // isStreaming indicates whether the response is a streaming type (e.g., text/event-stream).
 	streamWriter logging.StreamingLogWriter // streamWriter is a writer for handling streaming log entries.
 	chunkChannel chan []byte                // chunkChannel is a channel for asynchronously passing response chunks to the logger.
+	streamDone   chan struct{}              // streamDone signals when the streaming goroutine completes.
 	logger       logging.RequestLogger      // logger is the instance of the request logger service.
 	requestInfo  *RequestInfo               // requestInfo holds the details of the original request.
 	statusCode   int                        // statusCode stores the HTTP status code of the response.
@@ -107,9 +109,11 @@ func (w *ResponseWriterWrapper) WriteHeader(statusCode int) {
 		if err == nil {
 			w.streamWriter = streamWriter
 			w.chunkChannel = make(chan []byte, 100) // Buffered channel for async writes
+			doneChan := make(chan struct{})
+			w.streamDone = doneChan
 
 			// Start async chunk processor
-			go w.processStreamingChunks()
+			go w.processStreamingChunks(doneChan)
 
 			// Write status immediately
 			_ = streamWriter.WriteStatus(statusCode, w.headers)
@@ -167,7 +171,13 @@ func (w *ResponseWriterWrapper) detectStreaming(contentType string) bool {
 
 // processStreamingChunks runs in a separate goroutine to process response chunks from the chunkChannel.
 // It asynchronously writes each chunk to the streaming log writer.
-func (w *ResponseWriterWrapper) processStreamingChunks() {
+func (w *ResponseWriterWrapper) processStreamingChunks(done chan struct{}) {
+	if done == nil {
+		return
+	}
+
+	defer close(done)
+
 	if w.streamWriter == nil || w.chunkChannel == nil {
 		return
 	}
@@ -193,8 +203,15 @@ func (w *ResponseWriterWrapper) Finalize(c *gin.Context) error {
 			w.chunkChannel = nil
 		}
 
+		if w.streamDone != nil {
+			<-w.streamDone
+			w.streamDone = nil
+		}
+
 		if w.streamWriter != nil {
-			return w.streamWriter.Close()
+			err := w.streamWriter.Close()
+			w.streamWriter = nil
+			return err
 		}
 	} else {
 		// Capture final status code and headers if not already captured
@@ -240,6 +257,16 @@ func (w *ResponseWriterWrapper) Finalize(c *gin.Context) error {
 			}
 		}
 
+		var slicesAPIResponseError []*interfaces.ErrorMessage
+		apiResponseError, isExist := c.Get("API_RESPONSE_ERROR")
+		if isExist {
+			var ok bool
+			slicesAPIResponseError, ok = apiResponseError.([]*interfaces.ErrorMessage)
+			if !ok {
+				slicesAPIResponseError = nil
+			}
+		}
+
 		// Log complete non-streaming response
 		return w.logger.LogRequest(
 			w.requestInfo.URL,
@@ -251,6 +278,7 @@ func (w *ResponseWriterWrapper) Finalize(c *gin.Context) error {
 			w.body.Bytes(),
 			apiRequestBody,
 			apiResponseBody,
+			slicesAPIResponseError,
 		)
 	}
 

internal/api/server.go

@@ -9,17 +9,22 @@ import (
 	"errors"
 	"fmt"
 	"net/http"
+	"os"
+	"path/filepath"
 	"strings"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/api/handlers"
-	"github.com/luispater/CLIProxyAPI/internal/api/handlers/claude"
-	"github.com/luispater/CLIProxyAPI/internal/api/handlers/gemini"
-	"github.com/luispater/CLIProxyAPI/internal/api/handlers/openai"
-	"github.com/luispater/CLIProxyAPI/internal/api/middleware"
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/logging"
+	"github.com/luispater/CLIProxyAPI/v5/internal/api/handlers"
+	"github.com/luispater/CLIProxyAPI/v5/internal/api/handlers/claude"
+	"github.com/luispater/CLIProxyAPI/v5/internal/api/handlers/gemini"
+	managementHandlers "github.com/luispater/CLIProxyAPI/v5/internal/api/handlers/management"
+	"github.com/luispater/CLIProxyAPI/v5/internal/api/handlers/openai"
+	"github.com/luispater/CLIProxyAPI/v5/internal/api/middleware"
+	"github.com/luispater/CLIProxyAPI/v5/internal/client"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/logging"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 )
 
@@ -40,6 +45,12 @@ type Server struct {
 
 	// requestLogger is the request logger instance for dynamic configuration updates.
 	requestLogger *logging.FileRequestLogger
+
+	// configFilePath is the absolute path to the YAML config file for persistence.
+	configFilePath string
+
+	// management handler
+	mgmt *managementHandlers.Handler
 }
 
 // NewServer creates and initializes a new API server instance.
@@ -51,7 +62,7 @@ type Server struct {
 //
 // Returns:
 //   - *Server: A new server instance
-func NewServer(cfg *config.Config, cliClients []interfaces.Client) *Server {
+func NewServer(cfg *config.Config, cliClients []interfaces.Client, configFilePath string) *Server {
 	// Set gin mode
 	if !cfg.Debug {
 		gin.SetMode(gin.ReleaseMode)
@@ -65,18 +76,22 @@ func NewServer(cfg *config.Config, cliClients []interfaces.Client) *Server {
 	engine.Use(gin.Recovery())
 
 	// Add request logging middleware (positioned after recovery, before auth)
-	requestLogger := logging.NewFileRequestLogger(cfg.RequestLog, "logs")
+	// Resolve logs directory relative to the configuration file directory.
+	requestLogger := logging.NewFileRequestLogger(cfg.RequestLog, "logs", filepath.Dir(configFilePath))
 	engine.Use(middleware.RequestLoggingMiddleware(requestLogger))
 
 	engine.Use(corsMiddleware())
 
 	// Create server instance
 	s := &Server{
-		engine:        engine,
-		handlers:      handlers.NewBaseAPIHandlers(cliClients, cfg),
-		cfg:           cfg,
-		requestLogger: requestLogger,
+		engine:         engine,
+		handlers:       handlers.NewBaseAPIHandlers(cliClients, cfg),
+		cfg:            cfg,
+		requestLogger:  requestLogger,
+		configFilePath: configFilePath,
 	}
+	// Initialize management handler
+	s.mgmt = managementHandlers.NewHandler(cfg, configFilePath)
 
 	// Setup routes
 	s.setupRoutes()
@@ -97,6 +112,7 @@ func (s *Server) setupRoutes() {
 	geminiHandlers := gemini.NewGeminiAPIHandler(s.handlers)
 	geminiCLIHandlers := gemini.NewGeminiCLIAPIHandler(s.handlers)
 	claudeCodeHandlers := claude.NewClaudeCodeAPIHandler(s.handlers)
+	openaiResponsesHandlers := openai.NewOpenAIResponsesAPIHandler(s.handlers)
 
 	// OpenAI compatible API routes
 	v1 := s.engine.Group("/v1")
@@ -104,7 +120,9 @@ func (s *Server) setupRoutes() {
 	{
 		v1.GET("/models", s.unifiedModelsHandler(openaiHandlers, claudeCodeHandlers))
 		v1.POST("/chat/completions", openaiHandlers.ChatCompletions)
+		v1.POST("/completions", openaiHandlers.Completions)
 		v1.POST("/messages", claudeCodeHandlers.ClaudeMessages)
+		v1.POST("/responses", openaiResponsesHandlers.Responses)
 	}
 
 	// Gemini compatible API routes
@@ -123,11 +141,131 @@ func (s *Server) setupRoutes() {
 			"version": "1.0.0",
 			"endpoints": []string{
 				"POST /v1/chat/completions",
+				"POST /v1/completions",
 				"GET /v1/models",
 			},
 		})
 	})
 	s.engine.POST("/v1internal:method", geminiCLIHandlers.CLIHandler)
+
+	// OAuth callback endpoints (reuse main server port)
+	// These endpoints receive provider redirects and persist
+	// the short-lived code/state for the waiting goroutine.
+	s.engine.GET("/anthropic/callback", func(c *gin.Context) {
+		code := c.Query("code")
+		state := c.Query("state")
+		errStr := c.Query("error")
+		// Persist to a temporary file keyed by state
+		if state != "" {
+			file := fmt.Sprintf("%s/.oauth-anthropic-%s.oauth", s.cfg.AuthDir, state)
+			_ = os.WriteFile(file, []byte(fmt.Sprintf(`{"code":"%s","state":"%s","error":"%s"}`, code, state, errStr)), 0o600)
+		}
+		c.Header("Content-Type", "text/html; charset=utf-8")
+		c.String(http.StatusOK, "<html><body><h1>Authentication successful!</h1><p>You can close this window.</p></body></html>")
+	})
+
+	s.engine.GET("/codex/callback", func(c *gin.Context) {
+		code := c.Query("code")
+		state := c.Query("state")
+		errStr := c.Query("error")
+		if state != "" {
+			file := fmt.Sprintf("%s/.oauth-codex-%s.oauth", s.cfg.AuthDir, state)
+			_ = os.WriteFile(file, []byte(fmt.Sprintf(`{"code":"%s","state":"%s","error":"%s"}`, code, state, errStr)), 0o600)
+		}
+		c.Header("Content-Type", "text/html; charset=utf-8")
+		c.String(http.StatusOK, "<html><body><h1>Authentication successful!</h1><p>You can close this window.</p></body></html>")
+	})
+
+	s.engine.GET("/google/callback", func(c *gin.Context) {
+		code := c.Query("code")
+		state := c.Query("state")
+		errStr := c.Query("error")
+		if state != "" {
+			file := fmt.Sprintf("%s/.oauth-gemini-%s.oauth", s.cfg.AuthDir, state)
+			_ = os.WriteFile(file, []byte(fmt.Sprintf(`{"code":"%s","state":"%s","error":"%s"}`, code, state, errStr)), 0o600)
+		}
+		c.Header("Content-Type", "text/html; charset=utf-8")
+		c.String(http.StatusOK, "<html><body><h1>Authentication successful!</h1><p>You can close this window.</p></body></html>")
+	})
+
+	// Management API routes (delegated to management handlers)
+	// New logic: if remote-management-key is empty, do not expose any management endpoint (404).
+	if s.cfg.RemoteManagement.SecretKey != "" {
+		mgmt := s.engine.Group("/v0/management")
+		mgmt.Use(s.mgmt.Middleware())
+		{
+			mgmt.GET("/config", s.mgmt.GetConfig)
+
+			mgmt.GET("/debug", s.mgmt.GetDebug)
+			mgmt.PUT("/debug", s.mgmt.PutDebug)
+			mgmt.PATCH("/debug", s.mgmt.PutDebug)
+
+			mgmt.GET("/force-gpt-5-codex", s.mgmt.GetForceGPT5Codex)
+			mgmt.PUT("/force-gpt-5-codex", s.mgmt.PutForceGPT5Codex)
+			mgmt.PATCH("/force-gpt-5-codex", s.mgmt.PutForceGPT5Codex)
+
+			mgmt.GET("/proxy-url", s.mgmt.GetProxyURL)
+			mgmt.PUT("/proxy-url", s.mgmt.PutProxyURL)
+			mgmt.PATCH("/proxy-url", s.mgmt.PutProxyURL)
+			mgmt.DELETE("/proxy-url", s.mgmt.DeleteProxyURL)
+
+			mgmt.GET("/quota-exceeded/switch-project", s.mgmt.GetSwitchProject)
+			mgmt.PUT("/quota-exceeded/switch-project", s.mgmt.PutSwitchProject)
+			mgmt.PATCH("/quota-exceeded/switch-project", s.mgmt.PutSwitchProject)
+
+			mgmt.GET("/quota-exceeded/switch-preview-model", s.mgmt.GetSwitchPreviewModel)
+			mgmt.PUT("/quota-exceeded/switch-preview-model", s.mgmt.PutSwitchPreviewModel)
+			mgmt.PATCH("/quota-exceeded/switch-preview-model", s.mgmt.PutSwitchPreviewModel)
+
+			mgmt.GET("/api-keys", s.mgmt.GetAPIKeys)
+			mgmt.PUT("/api-keys", s.mgmt.PutAPIKeys)
+			mgmt.PATCH("/api-keys", s.mgmt.PatchAPIKeys)
+			mgmt.DELETE("/api-keys", s.mgmt.DeleteAPIKeys)
+
+			mgmt.GET("/generative-language-api-key", s.mgmt.GetGlKeys)
+			mgmt.PUT("/generative-language-api-key", s.mgmt.PutGlKeys)
+			mgmt.PATCH("/generative-language-api-key", s.mgmt.PatchGlKeys)
+			mgmt.DELETE("/generative-language-api-key", s.mgmt.DeleteGlKeys)
+
+			mgmt.GET("/request-log", s.mgmt.GetRequestLog)
+			mgmt.PUT("/request-log", s.mgmt.PutRequestLog)
+			mgmt.PATCH("/request-log", s.mgmt.PutRequestLog)
+
+			mgmt.GET("/request-retry", s.mgmt.GetRequestRetry)
+			mgmt.PUT("/request-retry", s.mgmt.PutRequestRetry)
+			mgmt.PATCH("/request-retry", s.mgmt.PutRequestRetry)
+
+			mgmt.GET("/allow-localhost-unauthenticated", s.mgmt.GetAllowLocalhost)
+			mgmt.PUT("/allow-localhost-unauthenticated", s.mgmt.PutAllowLocalhost)
+			mgmt.PATCH("/allow-localhost-unauthenticated", s.mgmt.PutAllowLocalhost)
+
+			mgmt.GET("/claude-api-key", s.mgmt.GetClaudeKeys)
+			mgmt.PUT("/claude-api-key", s.mgmt.PutClaudeKeys)
+			mgmt.PATCH("/claude-api-key", s.mgmt.PatchClaudeKey)
+			mgmt.DELETE("/claude-api-key", s.mgmt.DeleteClaudeKey)
+
+			mgmt.GET("/codex-api-key", s.mgmt.GetCodexKeys)
+			mgmt.PUT("/codex-api-key", s.mgmt.PutCodexKeys)
+			mgmt.PATCH("/codex-api-key", s.mgmt.PatchCodexKey)
+			mgmt.DELETE("/codex-api-key", s.mgmt.DeleteCodexKey)
+
+			mgmt.GET("/openai-compatibility", s.mgmt.GetOpenAICompat)
+			mgmt.PUT("/openai-compatibility", s.mgmt.PutOpenAICompat)
+			mgmt.PATCH("/openai-compatibility", s.mgmt.PatchOpenAICompat)
+			mgmt.DELETE("/openai-compatibility", s.mgmt.DeleteOpenAICompat)
+
+			mgmt.GET("/auth-files", s.mgmt.ListAuthFiles)
+			mgmt.GET("/auth-files/download", s.mgmt.DownloadAuthFile)
+			mgmt.POST("/auth-files", s.mgmt.UploadAuthFile)
+			mgmt.DELETE("/auth-files", s.mgmt.DeleteAuthFile)
+
+			mgmt.GET("/anthropic-auth-url", s.mgmt.RequestAnthropicToken)
+			mgmt.GET("/codex-auth-url", s.mgmt.RequestCodexToken)
+			mgmt.GET("/gemini-cli-auth-url", s.mgmt.RequestGeminiCLIToken)
+			mgmt.GET("/qwen-auth-url", s.mgmt.RequestQwenToken)
+			mgmt.GET("/get-auth-status", s.mgmt.GetAuthStatus)
+		}
+	}
 }
 
 // unifiedModelsHandler creates a unified handler for the /v1/models endpoint
@@ -140,10 +278,10 @@ func (s *Server) unifiedModelsHandler(openaiHandler *openai.OpenAIAPIHandler, cl
 
 		// Route to Claude handler if User-Agent starts with "claude-cli"
 		if strings.HasPrefix(userAgent, "claude-cli") {
-			log.Debugf("Routing /v1/models to Claude handler for User-Agent: %s", userAgent)
+			// log.Debugf("Routing /v1/models to Claude handler for User-Agent: %s", userAgent)
 			claudeHandler.ClaudeModels(c)
 		} else {
-			log.Debugf("Routing /v1/models to OpenAI handler for User-Agent: %s", userAgent)
+			// log.Debugf("Routing /v1/models to OpenAI handler for User-Agent: %s", userAgent)
 			openaiHandler.OpenAIModels(c)
 		}
 	}
@@ -194,7 +332,7 @@ func corsMiddleware() gin.HandlerFunc {
 	return func(c *gin.Context) {
 		c.Header("Access-Control-Allow-Origin", "*")
 		c.Header("Access-Control-Allow-Methods", "GET, POST, PUT, DELETE, OPTIONS")
-		c.Header("Access-Control-Allow-Headers", "Origin, Content-Type, Content-Length, Accept-Encoding, X-CSRF-Token, Authorization")
+		c.Header("Access-Control-Allow-Headers", "*")
 
 		if c.Request.Method == "OPTIONS" {
 			c.AbortWithStatus(http.StatusNoContent)
@@ -211,18 +349,72 @@ func corsMiddleware() gin.HandlerFunc {
 // Parameters:
 //   - clients: The new slice of AI service clients
 //   - cfg: The new application configuration
-func (s *Server) UpdateClients(clients []interfaces.Client, cfg *config.Config) {
+func (s *Server) UpdateClients(clients map[string]interfaces.Client, cfg *config.Config) {
+	clientSlice := s.clientsToSlice(clients)
 	// Update request logger enabled state if it has changed
 	if s.requestLogger != nil && s.cfg.RequestLog != cfg.RequestLog {
 		s.requestLogger.SetEnabled(cfg.RequestLog)
 		log.Debugf("request logging updated from %t to %t", s.cfg.RequestLog, cfg.RequestLog)
 	}
 
+	// Update log level dynamically when debug flag changes
+	if s.cfg.Debug != cfg.Debug {
+		util.SetLogLevel(cfg)
+		log.Debugf("debug mode updated from %t to %t", s.cfg.Debug, cfg.Debug)
+	}
+
 	s.cfg = cfg
-	s.handlers.UpdateClients(clients, cfg)
-	log.Infof("server clients and configuration updated: %d clients", len(clients))
+	s.handlers.UpdateClients(clientSlice, cfg)
+	if s.mgmt != nil {
+		s.mgmt.SetConfig(cfg)
+	}
+
+	// Count client types for detailed logging
+	authFiles := 0
+	glAPIKeyCount := 0
+	claudeAPIKeyCount := 0
+	codexAPIKeyCount := 0
+	openAICompatCount := 0
+
+	for _, c := range clientSlice {
+		switch cl := c.(type) {
+		case *client.GeminiCLIClient:
+			authFiles++
+		case *client.GeminiWebClient:
+			authFiles++
+		case *client.CodexClient:
+			if cl.GetAPIKey() == "" {
+				authFiles++
+			} else {
+				codexAPIKeyCount++
+			}
+		case *client.ClaudeClient:
+			if cl.GetAPIKey() == "" {
+				authFiles++
+			} else {
+				claudeAPIKeyCount++
+			}
+		case *client.QwenClient:
+			authFiles++
+		case *client.GeminiClient:
+			glAPIKeyCount++
+		case *client.OpenAICompatibilityClient:
+			openAICompatCount++
+		}
+	}
+
+	log.Infof("server clients and configuration updated: %d clients (%d auth files + %d GL API keys + %d Claude API keys + %d Codex keys + %d OpenAI-compat)",
+		len(clientSlice),
+		authFiles,
+		glAPIKeyCount,
+		claudeAPIKeyCount,
+		codexAPIKeyCount,
+		openAICompatCount,
+	)
 }
 
+// (management handlers moved to internal/api/handlers/management)
+
 // AuthMiddleware returns a Gin middleware handler that authenticates requests
 // using API keys. If no API keys are configured, it allows all requests.
 //
@@ -288,3 +480,11 @@ func AuthMiddleware(cfg *config.Config) gin.HandlerFunc {
 		c.Next()
 	}
 }
+
+func (s *Server) clientsToSlice(clientMap map[string]interfaces.Client) []interfaces.Client {
+	slice := make([]interfaces.Client, 0, len(clientMap))
+	for _, v := range clientMap {
+		slice = append(slice, v)
+	}
+	return slice
+}

internal/auth/claude/anthropic_auth.go

@@ -13,8 +13,8 @@ import (
 	"strings"
 	"time"
 
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	"github.com/luispater/CLIProxyAPI/internal/util"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 )
 

internal/auth/claude/token.go

@@ -7,7 +7,9 @@ import (
 	"encoding/json"
 	"fmt"
 	"os"
-	"path"
+	"path/filepath"
+
+	"github.com/luispater/CLIProxyAPI/v5/internal/misc"
 )
 
 // ClaudeTokenStorage stores OAuth2 token information for Anthropic Claude API authentication.
@@ -46,10 +48,11 @@ type ClaudeTokenStorage struct {
 // Returns:
 //   - error: An error if the operation fails, nil otherwise
 func (ts *ClaudeTokenStorage) SaveTokenToFile(authFilePath string) error {
+	misc.LogSavingCredentials(authFilePath)
 	ts.Type = "claude"
 
 	// Create directory structure if it doesn't exist
-	if err := os.MkdirAll(path.Dir(authFilePath), 0700); err != nil {
+	if err := os.MkdirAll(filepath.Dir(authFilePath), 0700); err != nil {
 		return fmt.Errorf("failed to create directory: %v", err)
 	}
 

internal/auth/codex/openai_auth.go

@@ -14,8 +14,8 @@ import (
 	"strings"
 	"time"
 
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	"github.com/luispater/CLIProxyAPI/internal/util"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 )
 

internal/auth/codex/token.go

@@ -7,7 +7,9 @@ import (
 	"encoding/json"
 	"fmt"
 	"os"
-	"path"
+	"path/filepath"
+
+	"github.com/luispater/CLIProxyAPI/v5/internal/misc"
 )
 
 // CodexTokenStorage stores OAuth2 token information for OpenAI Codex API authentication.
@@ -42,8 +44,9 @@ type CodexTokenStorage struct {
 // Returns:
 //   - error: An error if the operation fails, nil otherwise
 func (ts *CodexTokenStorage) SaveTokenToFile(authFilePath string) error {
+	misc.LogSavingCredentials(authFilePath)
 	ts.Type = "codex"
-	if err := os.MkdirAll(path.Dir(authFilePath), 0700); err != nil {
+	if err := os.MkdirAll(filepath.Dir(authFilePath), 0700); err != nil {
 		return fmt.Errorf("failed to create directory: %v", err)
 	}
 

internal/auth/gemini/gemini-web_token.go

@@ -0,0 +1,45 @@
+// Package gemini provides authentication and token management functionality
+// for Google's Gemini AI services. It handles OAuth2 token storage, serialization,
+// and retrieval for maintaining authenticated sessions with the Gemini API.
+package gemini
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+
+	"github.com/luispater/CLIProxyAPI/v5/internal/misc"
+	log "github.com/sirupsen/logrus"
+)
+
+// GeminiWebTokenStorage stores cookie information for Google Gemini Web authentication.
+type GeminiWebTokenStorage struct {
+	Secure1PSID   string `json:"secure_1psid"`
+	Secure1PSIDTS string `json:"secure_1psidts"`
+	Type          string `json:"type"`
+}
+
+// SaveTokenToFile serializes the Gemini Web token storage to a JSON file.
+func (ts *GeminiWebTokenStorage) SaveTokenToFile(authFilePath string) error {
+	misc.LogSavingCredentials(authFilePath)
+	ts.Type = "gemini-web"
+	if err := os.MkdirAll(filepath.Dir(authFilePath), 0700); err != nil {
+		return fmt.Errorf("failed to create directory: %v", err)
+	}
+
+	f, err := os.Create(authFilePath)
+	if err != nil {
+		return fmt.Errorf("failed to create token file: %w", err)
+	}
+	defer func() {
+		if errClose := f.Close(); errClose != nil {
+			log.Errorf("failed to close file: %v", errClose)
+		}
+	}()
+
+	if err = json.NewEncoder(f).Encode(ts); err != nil {
+		return fmt.Errorf("failed to write token to file: %w", err)
+	}
+	return nil
+}

internal/auth/gemini/gemini_auth.go

@@ -15,9 +15,10 @@ import (
 	"net/url"
 	"time"
 
-	"github.com/luispater/CLIProxyAPI/internal/auth/codex"
-	"github.com/luispater/CLIProxyAPI/internal/browser"
-	"github.com/luispater/CLIProxyAPI/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/codex"
+	"github.com/luispater/CLIProxyAPI/v5/internal/browser"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 	"github.com/tidwall/gjson"
 	"golang.org/x/net/proxy"
@@ -250,11 +251,13 @@ func (g *GeminiAuth) getTokenFromWeb(ctx context.Context, config *oauth2.Config,
 		// Check if browser is available
 		if !browser.IsAvailable() {
 			log.Warn("No browser available on this system")
+			util.PrintSSHTunnelInstructions(8085)
 			log.Infof("Please manually open this URL in your browser:\n\n%s\n", authURL)
 		} else {
 			if err := browser.OpenURL(authURL); err != nil {
 				authErr := codex.NewAuthenticationError(codex.ErrBrowserOpenFailed, err)
 				log.Warn(codex.GetUserFriendlyMessage(authErr))
+				util.PrintSSHTunnelInstructions(8085)
 				log.Infof("Please manually open this URL in your browser:\n\n%s\n", authURL)
 
 				// Log platform info for debugging
@@ -265,6 +268,7 @@ func (g *GeminiAuth) getTokenFromWeb(ctx context.Context, config *oauth2.Config,
 			}
 		}
 	} else {
+		util.PrintSSHTunnelInstructions(8085)
 		log.Infof("Please open this URL in your browser:\n\n%s\n", authURL)
 	}
 

internal/auth/gemini/gemini_token.go

@@ -7,8 +7,9 @@ import (
 	"encoding/json"
 	"fmt"
 	"os"
-	"path"
+	"path/filepath"
 
+	"github.com/luispater/CLIProxyAPI/v5/internal/misc"
 	log "github.com/sirupsen/logrus"
 )
 
@@ -45,8 +46,9 @@ type GeminiTokenStorage struct {
 // Returns:
 //   - error: An error if the operation fails, nil otherwise
 func (ts *GeminiTokenStorage) SaveTokenToFile(authFilePath string) error {
+	misc.LogSavingCredentials(authFilePath)
 	ts.Type = "gemini"
-	if err := os.MkdirAll(path.Dir(authFilePath), 0700); err != nil {
+	if err := os.MkdirAll(filepath.Dir(authFilePath), 0700); err != nil {
 		return fmt.Errorf("failed to create directory: %v", err)
 	}
 

internal/auth/qwen/qwen_auth.go

@@ -13,8 +13,8 @@ import (
 	"strings"
 	"time"
 
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	"github.com/luispater/CLIProxyAPI/internal/util"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 )
 

internal/auth/qwen/qwen_token.go

@@ -7,7 +7,9 @@ import (
 	"encoding/json"
 	"fmt"
 	"os"
-	"path"
+	"path/filepath"
+
+	"github.com/luispater/CLIProxyAPI/v5/internal/misc"
 )
 
 // QwenTokenStorage stores OAuth2 token information for Alibaba Qwen API authentication.
@@ -40,8 +42,9 @@ type QwenTokenStorage struct {
 // Returns:
 //   - error: An error if the operation fails, nil otherwise
 func (ts *QwenTokenStorage) SaveTokenToFile(authFilePath string) error {
+	misc.LogSavingCredentials(authFilePath)
 	ts.Type = "qwen"
-	if err := os.MkdirAll(path.Dir(authFilePath), 0700); err != nil {
+	if err := os.MkdirAll(filepath.Dir(authFilePath), 0700); err != nil {
 		return fmt.Errorf("failed to create directory: %v", err)
 	}
 

internal/browser/browser.go

@@ -21,7 +21,7 @@ import (
 // Returns:
 //   - An error if the URL cannot be opened, otherwise nil.
 func OpenURL(url string) error {
-	log.Debugf("Attempting to open URL in browser: %s", url)
+	log.Infof("Attempting to open URL in browser: %s", url)
 
 	// Try using the open-golang library first
 	err := open.Run(url)

internal/client/claude_client.go

@@ -16,16 +16,16 @@ import (
 	"time"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/auth"
-	"github.com/luispater/CLIProxyAPI/internal/auth/claude"
-	"github.com/luispater/CLIProxyAPI/internal/auth/empty"
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/misc"
-	"github.com/luispater/CLIProxyAPI/internal/registry"
-	"github.com/luispater/CLIProxyAPI/internal/translator/translator"
-	"github.com/luispater/CLIProxyAPI/internal/util"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/claude"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/empty"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/misc"
+	"github.com/luispater/CLIProxyAPI/v5/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/translator/translator"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
@@ -67,6 +67,7 @@ func NewClaudeClient(cfg *config.Config, ts *claude.ClaudeTokenStorage) *ClaudeC
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
 			tokenStorage:       ts,
+			isAvailable:        true,
 		},
 		claudeAuth:  claude.NewClaudeAuth(cfg),
 		apiKeyIndex: -1,
@@ -102,6 +103,7 @@ func NewClaudeClientWithKey(cfg *config.Config, apiKeyIndex int) *ClaudeClient {
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
 			tokenStorage:       &empty.EmptyStorage{},
+			isAvailable:        true,
 		},
 		claudeAuth:  claude.NewClaudeAuth(cfg),
 		apiKeyIndex: apiKeyIndex,
@@ -181,6 +183,7 @@ func (c *ClaudeClient) TokenStorage() auth.TokenStorage {
 //   - []byte: The response body.
 //   - *interfaces.ErrorMessage: An error message if the request fails.
 func (c *ClaudeClient) SendRawMessage(ctx context.Context, modelName string, rawJSON []byte, alt string) ([]byte, *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
 	handler := ctx.Value("handler").(interfaces.APIHandler)
 	handlerType := handler.HandlerType()
 	rawJSON = translator.Request(handlerType, c.Type(), modelName, rawJSON, false)
@@ -208,7 +211,7 @@ func (c *ClaudeClient) SendRawMessage(ctx context.Context, modelName string, raw
 	c.AddAPIResponseData(ctx, bodyBytes)
 
 	var param any
-	bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, bodyBytes, &param))
+	bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, originalRequestRawJSON, rawJSON, bodyBytes, &param))
 
 	return bodyBytes, nil
 }
@@ -226,6 +229,8 @@ func (c *ClaudeClient) SendRawMessage(ctx context.Context, modelName string, raw
 //   - <-chan []byte: A channel for receiving response data chunks.
 //   - <-chan *interfaces.ErrorMessage: A channel for receiving error messages.
 func (c *ClaudeClient) SendRawMessageStream(ctx context.Context, modelName string, rawJSON []byte, alt string) (<-chan []byte, <-chan *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
+
 	handler := ctx.Value("handler").(interfaces.APIHandler)
 	handlerType := handler.HandlerType()
 	rawJSON = translator.Request(handlerType, c.Type(), modelName, rawJSON, true)
@@ -275,7 +280,7 @@ func (c *ClaudeClient) SendRawMessageStream(ctx context.Context, modelName strin
 			var param any
 			for scanner.Scan() {
 				line := scanner.Bytes()
-				lines := translator.Response(handlerType, c.Type(), ctx, modelName, line, &param)
+				lines := translator.Response(handlerType, c.Type(), ctx, modelName, originalRequestRawJSON, rawJSON, line, &param)
 				for i := 0; i < len(lines); i++ {
 					dataChan <- []byte(lines[i])
 				}
@@ -328,8 +333,16 @@ func (c *ClaudeClient) SendRawTokenCount(_ context.Context, _ string, _ []byte,
 // Returns:
 //   - error: An error if the save operation fails, nil otherwise.
 func (c *ClaudeClient) SaveTokenToFile() error {
-	fileName := filepath.Join(c.cfg.AuthDir, fmt.Sprintf("claude-%s.json", c.tokenStorage.(*claude.ClaudeTokenStorage).Email))
-	return c.tokenStorage.SaveTokenToFile(fileName)
+	// API-key based clients don't have a file-backed token to persist.
+	if c.apiKeyIndex != -1 {
+		return nil
+	}
+	ts, ok := c.tokenStorage.(*claude.ClaudeTokenStorage)
+	if !ok || ts == nil || ts.Email == "" {
+		return nil
+	}
+	fileName := filepath.Join(c.cfg.AuthDir, fmt.Sprintf("claude-%s.json", ts.Email))
+	return ts.SaveTokenToFile(fileName)
 }
 
 // RefreshTokens refreshes the access tokens if they have expired.
@@ -343,6 +356,10 @@ func (c *ClaudeClient) SaveTokenToFile() error {
 //   - error: An error if the refresh operation fails, nil otherwise.
 func (c *ClaudeClient) RefreshTokens(ctx context.Context) error {
 	// Check if we have a valid refresh token
+	if c.apiKeyIndex != -1 {
+		return fmt.Errorf("no refresh token available")
+	}
+
 	if c.tokenStorage == nil || c.tokenStorage.(*claude.ClaudeTokenStorage).RefreshToken == "" {
 		return fmt.Errorf("no refresh token available")
 	}
@@ -535,7 +552,7 @@ func (c *ClaudeClient) GetEmail() string {
 	if ts, ok := c.tokenStorage.(*claude.ClaudeTokenStorage); ok {
 		return ts.Email
 	} else {
-		return ""
+		return c.cfg.ClaudeKey[c.apiKeyIndex].APIKey
 	}
 }
 
@@ -557,3 +574,22 @@ func (c *ClaudeClient) IsModelQuotaExceeded(model string) bool {
 	}
 	return false
 }
+
+// GetRequestMutex returns the mutex used to synchronize requests for this client.
+// This ensures that only one request is processed at a time for quota management.
+//
+// Returns:
+//   - *sync.Mutex: The mutex used for request synchronization
+func (c *ClaudeClient) GetRequestMutex() *sync.Mutex {
+	return nil
+}
+
+// IsAvailable returns true if the client is available for use.
+func (c *ClaudeClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *ClaudeClient) SetUnavailable() {
+	c.isAvailable = false
+}

internal/client/client.go

@@ -11,9 +11,9 @@ import (
 	"time"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/auth"
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	"github.com/luispater/CLIProxyAPI/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/registry"
 )
 
 // ClientBase provides a common base structure for all AI API clients.
@@ -41,6 +41,9 @@ type ClientBase struct {
 
 	// modelRegistry is the global model registry for tracking model availability.
 	modelRegistry *registry.ModelRegistry
+
+	// unavailable tracks whether the client is unavailable
+	isAvailable bool
 }
 
 // GetRequestMutex returns the mutex used to synchronize requests for this client.

internal/client/codex_client.go

@@ -17,30 +17,33 @@ import (
 
 	"github.com/gin-gonic/gin"
 	"github.com/google/uuid"
-	"github.com/luispater/CLIProxyAPI/internal/auth"
-	"github.com/luispater/CLIProxyAPI/internal/auth/codex"
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/registry"
-	"github.com/luispater/CLIProxyAPI/internal/translator/translator"
-	"github.com/luispater/CLIProxyAPI/internal/util"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/codex"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/empty"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/translator/translator"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
 )
 
 const (
-	chatGPTEndpoint = "https://chatgpt.com/backend-api"
+	chatGPTEndpoint = "https://chatgpt.com/backend-api/codex"
 )
 
 // CodexClient implements the Client interface for OpenAI API
 type CodexClient struct {
 	ClientBase
 	codexAuth *codex.CodexAuth
+	// apiKeyIndex is the index of the API key to use from the config, -1 if not using API keys
+	apiKeyIndex int
 }
 
-// NewCodexClient creates a new OpenAI client instance
+// NewCodexClient creates a new OpenAI client instance using token-based authentication
 //
 // Parameters:
 //   - cfg: The application configuration.
@@ -62,8 +65,10 @@ func NewCodexClient(cfg *config.Config, ts *codex.CodexTokenStorage) (*CodexClie
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
 			tokenStorage:       ts,
+			isAvailable:        true,
 		},
-		codexAuth: codex.NewCodexAuth(cfg),
+		codexAuth:   codex.NewCodexAuth(cfg),
+		apiKeyIndex: -1,
 	}
 
 	// Initialize model registry and register OpenAI models
@@ -73,6 +78,42 @@ func NewCodexClient(cfg *config.Config, ts *codex.CodexTokenStorage) (*CodexClie
 	return client, nil
 }
 
+// NewCodexClientWithKey creates a new Codex client instance using API key authentication.
+// It initializes the client with the provided configuration and selects the API key
+// at the specified index from the configuration.
+//
+// Parameters:
+//   - cfg: The application configuration.
+//   - apiKeyIndex: The index of the API key to use from the configuration.
+//
+// Returns:
+//   - *CodexClient: A new Codex client instance.
+func NewCodexClientWithKey(cfg *config.Config, apiKeyIndex int) *CodexClient {
+	httpClient := util.SetProxy(cfg, &http.Client{})
+
+	// Generate unique client ID for API key client
+	clientID := fmt.Sprintf("codex-apikey-%d-%d", apiKeyIndex, time.Now().UnixNano())
+
+	client := &CodexClient{
+		ClientBase: ClientBase{
+			RequestMutex:       &sync.Mutex{},
+			httpClient:         httpClient,
+			cfg:                cfg,
+			modelQuotaExceeded: make(map[string]*time.Time),
+			tokenStorage:       &empty.EmptyStorage{},
+			isAvailable:        true,
+		},
+		codexAuth:   codex.NewCodexAuth(cfg),
+		apiKeyIndex: apiKeyIndex,
+	}
+
+	// Initialize model registry and register OpenAI models
+	client.InitializeModelRegistry(clientID)
+	client.RegisterModels("codex", registry.GetOpenAIModels())
+
+	return client
+}
+
 // Type returns the client type
 func (c *CodexClient) Type() string {
 	return CODEX
@@ -93,14 +134,29 @@ func (c *CodexClient) Provider() string {
 func (c *CodexClient) CanProvideModel(modelName string) bool {
 	models := []string{
 		"gpt-5",
-		"gpt-5-mini",
-		"gpt-5-nano",
+		"gpt-5-minimal",
+		"gpt-5-low",
+		"gpt-5-medium",
 		"gpt-5-high",
+		"gpt-5-codex",
+		"gpt-5-codex-low",
+		"gpt-5-codex-medium",
+		"gpt-5-codex-high",
 		"codex-mini-latest",
 	}
 	return util.InArray(models, modelName)
 }
 
+// GetAPIKey returns the API key for Codex API requests.
+// If an API key index is specified, it returns the corresponding key from the configuration.
+// Otherwise, it returns an empty string, indicating token-based authentication should be used.
+func (c *CodexClient) GetAPIKey() string {
+	if c.apiKeyIndex != -1 {
+		return c.cfg.CodexKey[c.apiKeyIndex].APIKey
+	}
+	return ""
+}
+
 // GetUserAgent returns the user agent string for OpenAI API requests
 func (c *CodexClient) GetUserAgent() string {
 	return "codex-cli"
@@ -123,11 +179,13 @@ func (c *CodexClient) TokenStorage() auth.TokenStorage {
 //   - []byte: The response body.
 //   - *interfaces.ErrorMessage: An error message if the request fails.
 func (c *CodexClient) SendRawMessage(ctx context.Context, modelName string, rawJSON []byte, alt string) ([]byte, *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
+
 	handler := ctx.Value("handler").(interfaces.APIHandler)
 	handlerType := handler.HandlerType()
 	rawJSON = translator.Request(handlerType, c.Type(), modelName, rawJSON, false)
 
-	respBody, err := c.APIRequest(ctx, modelName, "/codex/responses", rawJSON, alt, false)
+	respBody, err := c.APIRequest(ctx, modelName, "/responses", rawJSON, alt, false)
 	if err != nil {
 		if err.StatusCode == 429 {
 			now := time.Now()
@@ -149,7 +207,7 @@ func (c *CodexClient) SendRawMessage(ctx context.Context, modelName string, rawJ
 	c.AddAPIResponseData(ctx, bodyBytes)
 
 	var param any
-	bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, bodyBytes, &param))
+	bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, originalRequestRawJSON, rawJSON, bodyBytes, &param))
 
 	return bodyBytes, nil
 
@@ -167,6 +225,8 @@ func (c *CodexClient) SendRawMessage(ctx context.Context, modelName string, rawJ
 //   - <-chan []byte: A channel for receiving response data chunks.
 //   - <-chan *interfaces.ErrorMessage: A channel for receiving error messages.
 func (c *CodexClient) SendRawMessageStream(ctx context.Context, modelName string, rawJSON []byte, alt string) (<-chan []byte, <-chan *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
+
 	handler := ctx.Value("handler").(interfaces.APIHandler)
 	handlerType := handler.HandlerType()
 	rawJSON = translator.Request(handlerType, c.Type(), modelName, rawJSON, true)
@@ -192,7 +252,7 @@ func (c *CodexClient) SendRawMessageStream(ctx context.Context, modelName string
 		}
 
 		var err *interfaces.ErrorMessage
-		stream, err = c.APIRequest(ctx, modelName, "/codex/responses", rawJSON, alt, true)
+		stream, err = c.APIRequest(ctx, modelName, "/responses", rawJSON, alt, true)
 		if err != nil {
 			if err.StatusCode == 429 {
 				now := time.Now()
@@ -217,7 +277,7 @@ func (c *CodexClient) SendRawMessageStream(ctx context.Context, modelName string
 			var param any
 			for scanner.Scan() {
 				line := scanner.Bytes()
-				lines := translator.Response(handlerType, c.Type(), ctx, modelName, line, &param)
+				lines := translator.Response(handlerType, c.Type(), ctx, modelName, originalRequestRawJSON, rawJSON, line, &param)
 				for i := 0; i < len(lines); i++ {
 					dataChan <- []byte(lines[i])
 				}
@@ -266,8 +326,16 @@ func (c *CodexClient) SendRawTokenCount(_ context.Context, _ string, _ []byte, _
 // Returns:
 //   - error: An error if the save operation fails, nil otherwise.
 func (c *CodexClient) SaveTokenToFile() error {
-	fileName := filepath.Join(c.cfg.AuthDir, fmt.Sprintf("codex-%s.json", c.tokenStorage.(*codex.CodexTokenStorage).Email))
-	return c.tokenStorage.SaveTokenToFile(fileName)
+	// API-key based clients don't have a file-backed token to persist.
+	if c.apiKeyIndex != -1 {
+		return nil
+	}
+	ts, ok := c.tokenStorage.(*codex.CodexTokenStorage)
+	if !ok || ts == nil || ts.Email == "" {
+		return nil
+	}
+	fileName := filepath.Join(c.cfg.AuthDir, fmt.Sprintf("codex-%s.json", ts.Email))
+	return ts.SaveTokenToFile(fileName)
 }
 
 // RefreshTokens refreshes the access tokens if needed
@@ -278,6 +346,11 @@ func (c *CodexClient) SaveTokenToFile() error {
 // Returns:
 //   - error: An error if the refresh operation fails, nil otherwise.
 func (c *CodexClient) RefreshTokens(ctx context.Context) error {
+	// Check if we have a valid refresh token
+	if c.apiKeyIndex != -1 {
+		return fmt.Errorf("no refresh token available")
+	}
+
 	if c.tokenStorage == nil || c.tokenStorage.(*codex.CodexTokenStorage).RefreshToken == "" {
 		return fmt.Errorf("no refresh token available")
 	}
@@ -344,21 +417,52 @@ func (c *CodexClient) APIRequest(ctx context.Context, modelName, endpoint string
 	// Stream must be set to true
 	jsonBody, _ = sjson.SetBytes(jsonBody, "stream", true)
 
-	if util.InArray([]string{"gpt-5-nano", "gpt-5-mini", "gpt-5", "gpt-5-high"}, modelName) {
+	if util.InArray([]string{"gpt-5-minimal", "gpt-5-low", "gpt-5-medium", "gpt-5-high"}, modelName) {
 		jsonBody, _ = sjson.SetBytes(jsonBody, "model", "gpt-5")
 		switch modelName {
-		case "gpt-5-nano":
+		case "gpt-5-minimal":
 			jsonBody, _ = sjson.SetBytes(jsonBody, "reasoning.effort", "minimal")
-		case "gpt-5-mini":
+		case "gpt-5-low":
 			jsonBody, _ = sjson.SetBytes(jsonBody, "reasoning.effort", "low")
-		case "gpt-5":
+		case "gpt-5-medium":
 			jsonBody, _ = sjson.SetBytes(jsonBody, "reasoning.effort", "medium")
 		case "gpt-5-high":
 			jsonBody, _ = sjson.SetBytes(jsonBody, "reasoning.effort", "high")
 		}
+	} else if util.InArray([]string{"gpt-5-codex", "gpt-5-codex-low", "gpt-5-codex-medium", "gpt-5-codex-high"}, modelName) {
+		jsonBody, _ = sjson.SetBytes(jsonBody, "model", "gpt-5-codex")
+		switch modelName {
+		case "gpt-5-codex":
+			jsonBody, _ = sjson.SetBytes(jsonBody, "reasoning.effort", "medium")
+		case "gpt-5-codex-low":
+			jsonBody, _ = sjson.SetBytes(jsonBody, "reasoning.effort", "low")
+		case "gpt-5-codex-medium":
+			jsonBody, _ = sjson.SetBytes(jsonBody, "reasoning.effort", "medium")
+		case "gpt-5-codex-high":
+			jsonBody, _ = sjson.SetBytes(jsonBody, "reasoning.effort", "high")
+		}
+	} else if c.cfg.ForceGPT5Codex {
+		if gjson.GetBytes(jsonBody, "model").String() == "gpt-5" {
+			if gjson.GetBytes(jsonBody, "reasoning.effort").String() == "minimal" {
+				jsonBody, _ = sjson.SetBytes(jsonBody, "reasoning.effort", "low")
+			}
+			jsonBody, _ = sjson.SetBytes(jsonBody, "model", "gpt-5-codex")
+		}
 	}
 
 	url := fmt.Sprintf("%s%s", chatGPTEndpoint, endpoint)
+	accessToken := ""
+
+	if c.apiKeyIndex != -1 {
+		// Using API key authentication - use configured base URL if provided
+		if c.cfg.CodexKey[c.apiKeyIndex].BaseURL != "" {
+			url = fmt.Sprintf("%s%s", c.cfg.CodexKey[c.apiKeyIndex].BaseURL, endpoint)
+		}
+		accessToken = c.cfg.CodexKey[c.apiKeyIndex].APIKey
+	} else {
+		// Using OAuth token authentication - use ChatGPT endpoint
+		accessToken = c.tokenStorage.(*codex.CodexTokenStorage).AccessToken
+	}
 
 	// log.Debug(string(jsonBody))
 	// log.Debug(url)
@@ -376,9 +480,17 @@ func (c *CodexClient) APIRequest(ctx context.Context, modelName, endpoint string
 	req.Header.Set("Openai-Beta", "responses=experimental")
 	req.Header.Set("Session_id", sessionID)
 	req.Header.Set("Accept", "text/event-stream")
-	req.Header.Set("Chatgpt-Account-Id", c.tokenStorage.(*codex.CodexTokenStorage).AccountID)
-	req.Header.Set("Originator", "codex_cli_rs")
-	req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", c.tokenStorage.(*codex.CodexTokenStorage).AccessToken))
+	req.Header.Set("Connection", "Keep-Alive")
+
+	if c.apiKeyIndex != -1 {
+		// Using API key authentication
+		req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", accessToken))
+	} else {
+		// Using OAuth token authentication - include ChatGPT specific headers
+		req.Header.Set("Chatgpt-Account-Id", c.tokenStorage.(*codex.CodexTokenStorage).AccountID)
+		req.Header.Set("Originator", "codex_cli_rs")
+		req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", accessToken))
+	}
 
 	if c.cfg.RequestLog {
 		if ginContext, ok := ctx.Value("gin").(*gin.Context); ok {
@@ -386,7 +498,11 @@ func (c *CodexClient) APIRequest(ctx context.Context, modelName, endpoint string
 		}
 	}
 
-	log.Debugf("Use ChatGPT account %s for model %s", c.GetEmail(), modelName)
+	if c.apiKeyIndex != -1 {
+		log.Debugf("Use Codex API key %s for model %s", util.HideAPIKey(c.cfg.CodexKey[c.apiKeyIndex].APIKey), modelName)
+	} else {
+		log.Debugf("Use ChatGPT account %s for model %s", c.GetEmail(), modelName)
+	}
 
 	resp, err := c.httpClient.Do(req)
 	if err != nil {
@@ -408,7 +524,11 @@ func (c *CodexClient) APIRequest(ctx context.Context, modelName, endpoint string
 }
 
 // GetEmail returns the email associated with the client's token storage.
+// If the client is using API key authentication, it returns the API key.
 func (c *CodexClient) GetEmail() string {
+	if c.apiKeyIndex != -1 {
+		return c.cfg.CodexKey[c.apiKeyIndex].APIKey
+	}
 	return c.tokenStorage.(*codex.CodexTokenStorage).Email
 }
 
@@ -430,3 +550,22 @@ func (c *CodexClient) IsModelQuotaExceeded(model string) bool {
 	}
 	return false
 }
+
+// GetRequestMutex returns the mutex used to synchronize requests for this client.
+// This ensures that only one request is processed at a time for quota management.
+//
+// Returns:
+//   - *sync.Mutex: The mutex used for request synchronization
+func (c *CodexClient) GetRequestMutex() *sync.Mutex {
+	return nil
+}
+
+// IsAvailable returns true if the client is available for use.
+func (c *CodexClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *CodexClient) SetUnavailable() {
+	c.isAvailable = false
+}

internal/client/gemini-cli_client.go

@@ -18,13 +18,13 @@ import (
 	"time"
 
 	"github.com/gin-gonic/gin"
-	geminiAuth "github.com/luispater/CLIProxyAPI/internal/auth/gemini"
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/registry"
-	"github.com/luispater/CLIProxyAPI/internal/translator/translator"
-	"github.com/luispater/CLIProxyAPI/internal/util"
+	geminiAuth "github.com/luispater/CLIProxyAPI/v5/internal/auth/gemini"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/translator/translator"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
@@ -38,8 +38,9 @@ const (
 
 var (
 	previewModels = map[string][]string{
-		"gemini-2.5-pro":   {"gemini-2.5-pro-preview-05-06", "gemini-2.5-pro-preview-06-05"},
-		"gemini-2.5-flash": {"gemini-2.5-flash-preview-04-17", "gemini-2.5-flash-preview-05-20"},
+		"gemini-2.5-pro":        {"gemini-2.5-pro-preview-05-06", "gemini-2.5-pro-preview-06-05"},
+		"gemini-2.5-flash":      {"gemini-2.5-flash-preview-04-17", "gemini-2.5-flash-preview-05-20"},
+		"gemini-2.5-flash-lite": {"gemini-2.5-flash-lite-preview-06-17"},
 	}
 )
 
@@ -68,6 +69,7 @@ func NewGeminiCLIClient(httpClient *http.Client, ts *geminiAuth.GeminiTokenStora
 			cfg:                cfg,
 			tokenStorage:       ts,
 			modelQuotaExceeded: make(map[string]*time.Time),
+			isAvailable:        true,
 		},
 	}
 
@@ -99,6 +101,7 @@ func (c *GeminiCLIClient) CanProvideModel(modelName string) bool {
 	models := []string{
 		"gemini-2.5-pro",
 		"gemini-2.5-flash",
+		"gemini-2.5-flash-lite",
 	}
 	return util.InArray(models, modelName)
 }
@@ -407,6 +410,7 @@ func (c *GeminiCLIClient) APIRequest(ctx context.Context, modelName, endpoint st
 //   - []byte: The response body.
 //   - *interfaces.ErrorMessage: An error message if the request fails.
 func (c *GeminiCLIClient) SendRawTokenCount(ctx context.Context, modelName string, rawJSON []byte, alt string) ([]byte, *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
 	for {
 		if c.isModelQuotaExceeded(modelName) {
 			if c.cfg.QuotaExceeded.SwitchPreviewModel {
@@ -414,6 +418,7 @@ func (c *GeminiCLIClient) SendRawTokenCount(ctx context.Context, modelName strin
 				if newModelName != "" {
 					log.Debugf("Model %s is quota exceeded. Switch to preview model %s", modelName, newModelName)
 					rawJSON, _ = sjson.SetBytes(rawJSON, "model", newModelName)
+					modelName = newModelName
 					continue
 				}
 			}
@@ -453,7 +458,7 @@ func (c *GeminiCLIClient) SendRawTokenCount(ctx context.Context, modelName strin
 
 		c.AddAPIResponseData(ctx, bodyBytes)
 		var param any
-		bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, bodyBytes, &param))
+		bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, originalRequestRawJSON, rawJSON, bodyBytes, &param))
 
 		return bodyBytes, nil
 	}
@@ -471,6 +476,8 @@ func (c *GeminiCLIClient) SendRawTokenCount(ctx context.Context, modelName strin
 //   - []byte: The response body.
 //   - *interfaces.ErrorMessage: An error message if the request fails.
 func (c *GeminiCLIClient) SendRawMessage(ctx context.Context, modelName string, rawJSON []byte, alt string) ([]byte, *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
+
 	handler := ctx.Value("handler").(interfaces.APIHandler)
 	handlerType := handler.HandlerType()
 	rawJSON = translator.Request(handlerType, c.Type(), modelName, rawJSON, false)
@@ -484,6 +491,7 @@ func (c *GeminiCLIClient) SendRawMessage(ctx context.Context, modelName string,
 				if newModelName != "" {
 					log.Debugf("Model %s is quota exceeded. Switch to preview model %s", modelName, newModelName)
 					rawJSON, _ = sjson.SetBytes(rawJSON, "model", newModelName)
+					modelName = newModelName
 					continue
 				}
 			}
@@ -519,7 +527,7 @@ func (c *GeminiCLIClient) SendRawMessage(ctx context.Context, modelName string,
 
 		newCtx := context.WithValue(ctx, "alt", alt)
 		var param any
-		bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), newCtx, modelName, bodyBytes, &param))
+		bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), newCtx, modelName, originalRequestRawJSON, rawJSON, bodyBytes, &param))
 
 		return bodyBytes, nil
 	}
@@ -537,6 +545,8 @@ func (c *GeminiCLIClient) SendRawMessage(ctx context.Context, modelName string,
 //   - <-chan []byte: A channel for receiving response data chunks.
 //   - <-chan *interfaces.ErrorMessage: A channel for receiving error messages.
 func (c *GeminiCLIClient) SendRawMessageStream(ctx context.Context, modelName string, rawJSON []byte, alt string) (<-chan []byte, <-chan *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
+
 	handler := ctx.Value("handler").(interfaces.APIHandler)
 	handlerType := handler.HandlerType()
 	rawJSON = translator.Request(handlerType, c.Type(), modelName, rawJSON, true)
@@ -563,6 +573,7 @@ func (c *GeminiCLIClient) SendRawMessageStream(ctx context.Context, modelName st
 					if newModelName != "" {
 						log.Debugf("Model %s is quota exceeded. Switch to preview model %s", modelName, newModelName)
 						rawJSON, _ = sjson.SetBytes(rawJSON, "model", newModelName)
+						modelName = newModelName
 						continue
 					}
 				}
@@ -608,7 +619,7 @@ func (c *GeminiCLIClient) SendRawMessageStream(ctx context.Context, modelName st
 				for scanner.Scan() {
 					line := scanner.Bytes()
 					if bytes.HasPrefix(line, dataTag) {
-						lines := translator.Response(handlerType, c.Type(), newCtx, modelName, line[6:], &param)
+						lines := translator.Response(handlerType, c.Type(), newCtx, modelName, originalRequestRawJSON, rawJSON, line[6:], &param)
 						for i := 0; i < len(lines); i++ {
 							dataChan <- []byte(lines[i])
 						}
@@ -640,7 +651,7 @@ func (c *GeminiCLIClient) SendRawMessageStream(ctx context.Context, modelName st
 			}
 
 			if translator.NeedConvert(handlerType, c.Type()) {
-				lines := translator.Response(handlerType, c.Type(), newCtx, modelName, data, &param)
+				lines := translator.Response(handlerType, c.Type(), newCtx, modelName, originalRequestRawJSON, rawJSON, data, &param)
 				for i := 0; i < len(lines); i++ {
 					dataChan <- []byte(lines[i])
 				}
@@ -651,7 +662,7 @@ func (c *GeminiCLIClient) SendRawMessageStream(ctx context.Context, modelName st
 		}
 
 		if translator.NeedConvert(handlerType, c.Type()) {
-			lines := translator.Response(handlerType, c.Type(), ctx, modelName, []byte("[DONE]"), &param)
+			lines := translator.Response(handlerType, c.Type(), ctx, modelName, rawJSON, originalRequestRawJSON, []byte("[DONE]"), &param)
 			for i := 0; i < len(lines); i++ {
 				dataChan <- []byte(lines[i])
 			}
@@ -820,7 +831,6 @@ func (c *GeminiCLIClient) GetProjectList(ctx context.Context) (*interfaces.GCPPr
 //   - error: An error if the save operation fails, nil otherwise.
 func (c *GeminiCLIClient) SaveTokenToFile() error {
 	fileName := filepath.Join(c.cfg.AuthDir, fmt.Sprintf("%s-%s.json", c.tokenStorage.(*geminiAuth.GeminiTokenStorage).Email, c.tokenStorage.(*geminiAuth.GeminiTokenStorage).ProjectID))
-	log.Infof("Saving credentials to %s", fileName)
 	return c.tokenStorage.SaveTokenToFile(fileName)
 }
 
@@ -851,3 +861,28 @@ func (c *GeminiCLIClient) GetUserAgent() string {
 	// return fmt.Sprintf("GeminiCLI/%s (%s; %s)", pluginVersion, runtime.GOOS, runtime.GOARCH)
 	return "google-api-nodejs-client/9.15.1"
 }
+
+// GetRequestMutex returns the mutex used to synchronize requests for this client.
+// This ensures that only one request is processed at a time for quota management.
+//
+// Returns:
+//   - *sync.Mutex: The mutex used for request synchronization
+func (c *GeminiCLIClient) GetRequestMutex() *sync.Mutex {
+	return nil
+}
+
+// RefreshTokens is not applicable for Gemini CLI clients as they use API keys.
+func (c *GeminiCLIClient) RefreshTokens(ctx context.Context) error {
+	// API keys don't need refreshing
+	return nil
+}
+
+// IsAvailable returns true if the client is available for use.
+func (c *GeminiCLIClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *GeminiCLIClient) SetUnavailable() {
+	c.isAvailable = false
+}

internal/client/gemini-web/auth.go

@@ -0,0 +1,228 @@
+package geminiwebapi
+
+import (
+	"crypto/tls"
+	"errors"
+	"io"
+	"net/http"
+	"net/http/cookiejar"
+	"net/url"
+	"os"
+	"path/filepath"
+	"regexp"
+	"strings"
+	"time"
+)
+
+type httpOptions struct {
+	ProxyURL        string
+	Insecure        bool
+	FollowRedirects bool
+}
+
+func newHTTPClient(opts httpOptions) *http.Client {
+	transport := &http.Transport{}
+	if opts.ProxyURL != "" {
+		if pu, err := url.Parse(opts.ProxyURL); err == nil {
+			transport.Proxy = http.ProxyURL(pu)
+		}
+	}
+	if opts.Insecure {
+		transport.TLSClientConfig = &tls.Config{InsecureSkipVerify: true}
+	}
+	jar, _ := cookiejar.New(nil)
+	client := &http.Client{Transport: transport, Timeout: 60 * time.Second, Jar: jar}
+	if !opts.FollowRedirects {
+		client.CheckRedirect = func(req *http.Request, via []*http.Request) error {
+			return http.ErrUseLastResponse
+		}
+	}
+	return client
+}
+
+func applyHeaders(req *http.Request, headers http.Header) {
+	for k, v := range headers {
+		for _, vv := range v {
+			req.Header.Add(k, vv)
+		}
+	}
+}
+
+func applyCookies(req *http.Request, cookies map[string]string) {
+	for k, v := range cookies {
+		req.AddCookie(&http.Cookie{Name: k, Value: v})
+	}
+}
+
+func sendInitRequest(cookies map[string]string, proxy string, insecure bool) (*http.Response, map[string]string, error) {
+	client := newHTTPClient(httpOptions{ProxyURL: proxy, Insecure: insecure, FollowRedirects: true})
+	req, _ := http.NewRequest(http.MethodGet, EndpointInit, nil)
+	applyHeaders(req, HeadersGemini)
+	applyCookies(req, cookies)
+	resp, err := client.Do(req)
+	if err != nil {
+		return nil, nil, err
+	}
+	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
+		return resp, nil, &AuthError{Msg: resp.Status}
+	}
+	outCookies := map[string]string{}
+	for _, c := range resp.Cookies() {
+		outCookies[c.Name] = c.Value
+	}
+	for k, v := range cookies {
+		outCookies[k] = v
+	}
+	return resp, outCookies, nil
+}
+
+func getAccessToken(baseCookies map[string]string, proxy string, verbose bool, insecure bool) (string, map[string]string, error) {
+	// Warm-up google.com to gain extra cookies (NID, etc.) and capture them.
+	extraCookies := map[string]string{}
+	{
+		client := newHTTPClient(httpOptions{ProxyURL: proxy, Insecure: insecure, FollowRedirects: true})
+		req, _ := http.NewRequest(http.MethodGet, EndpointGoogle, nil)
+		resp, _ := client.Do(req)
+		if resp != nil {
+			if u, err := url.Parse(EndpointGoogle); err == nil {
+				for _, c := range client.Jar.Cookies(u) {
+					extraCookies[c.Name] = c.Value
+				}
+			}
+			_ = resp.Body.Close()
+		}
+	}
+
+	trySets := make([]map[string]string, 0, 8)
+
+	if v1, ok1 := baseCookies["__Secure-1PSID"]; ok1 {
+		if v2, ok2 := baseCookies["__Secure-1PSIDTS"]; ok2 {
+			merged := map[string]string{"__Secure-1PSID": v1, "__Secure-1PSIDTS": v2}
+			if nid, ok := baseCookies["NID"]; ok {
+				merged["NID"] = nid
+			}
+			trySets = append(trySets, merged)
+		} else if verbose {
+			Debug("Skipping base cookies: __Secure-1PSIDTS missing")
+		}
+	}
+
+	cacheDir := "temp"
+	_ = os.MkdirAll(cacheDir, 0o755)
+	if v1, ok1 := baseCookies["__Secure-1PSID"]; ok1 {
+		cacheFile := filepath.Join(cacheDir, ".cached_1psidts_"+v1+".txt")
+		if b, err := os.ReadFile(cacheFile); err == nil {
+			cv := strings.TrimSpace(string(b))
+			if cv != "" {
+				merged := map[string]string{"__Secure-1PSID": v1, "__Secure-1PSIDTS": cv}
+				trySets = append(trySets, merged)
+			}
+		}
+	}
+
+	if len(extraCookies) > 0 {
+		trySets = append(trySets, extraCookies)
+	}
+
+	reToken := regexp.MustCompile(`"SNlM0e":"([^"]+)"`)
+
+	for _, cookies := range trySets {
+		resp, mergedCookies, err := sendInitRequest(cookies, proxy, insecure)
+		if err != nil {
+			if verbose {
+				Warning("Failed init request: %v", err)
+			}
+			continue
+		}
+		body, err := io.ReadAll(resp.Body)
+		_ = resp.Body.Close()
+		if err != nil {
+			return "", nil, err
+		}
+		matches := reToken.FindStringSubmatch(string(body))
+		if len(matches) >= 2 {
+			token := matches[1]
+			if verbose {
+				Success("Gemini access token acquired.")
+			}
+			return token, mergedCookies, nil
+		}
+	}
+	return "", nil, &AuthError{Msg: "Failed to retrieve token."}
+}
+
+// rotate1psidts refreshes __Secure-1PSIDTS and caches it locally.
+func rotate1psidts(cookies map[string]string, proxy string, insecure bool) (string, error) {
+	psid, ok := cookies["__Secure-1PSID"]
+	if !ok {
+		return "", &AuthError{Msg: "__Secure-1PSID missing"}
+	}
+
+	cacheDir := "temp"
+	_ = os.MkdirAll(cacheDir, 0o755)
+	cacheFile := filepath.Join(cacheDir, ".cached_1psidts_"+psid+".txt")
+
+	if st, err := os.Stat(cacheFile); err == nil {
+		if time.Since(st.ModTime()) <= time.Minute {
+			if b, err := os.ReadFile(cacheFile); err == nil {
+				v := strings.TrimSpace(string(b))
+				if v != "" {
+					return v, nil
+				}
+			}
+		}
+	}
+
+	tr := &http.Transport{}
+	if proxy != "" {
+		if pu, err := url.Parse(proxy); err == nil {
+			tr.Proxy = http.ProxyURL(pu)
+		}
+	}
+	if insecure {
+		tr.TLSClientConfig = &tls.Config{InsecureSkipVerify: true}
+	}
+	client := &http.Client{Transport: tr, Timeout: 60 * time.Second}
+
+	req, _ := http.NewRequest(http.MethodPost, EndpointRotateCookies, io.NopCloser(stringsReader("[000,\"-0000000000000000000\"]")))
+	applyHeaders(req, HeadersRotateCookies)
+	applyCookies(req, cookies)
+
+	resp, err := client.Do(req)
+	if err != nil {
+		return "", err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode == http.StatusUnauthorized {
+		return "", &AuthError{Msg: "unauthorized"}
+	}
+	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
+		return "", errors.New(resp.Status)
+	}
+
+	for _, c := range resp.Cookies() {
+		if c.Name == "__Secure-1PSIDTS" {
+			_ = os.WriteFile(cacheFile, []byte(c.Value), 0o644)
+			return c.Value, nil
+		}
+	}
+	return "", nil
+}
+
+// Minimal reader helpers to avoid importing strings everywhere.
+type constReader struct {
+	s string
+	i int
+}
+
+func (r *constReader) Read(p []byte) (int, error) {
+	if r.i >= len(r.s) {
+		return 0, io.EOF
+	}
+	n := copy(p, r.s[r.i:])
+	r.i += n
+	return n, nil
+}
+
+func stringsReader(s string) io.Reader { return &constReader{s: s} }

internal/client/gemini-web/client.go

@@ -0,0 +1,797 @@
+package geminiwebapi
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"regexp"
+	"strings"
+	"sync"
+	"time"
+)
+
+// GeminiClient is the async http client interface (Go port)
+type GeminiClient struct {
+	Cookies         map[string]string
+	Proxy           string
+	Running         bool
+	httpClient      *http.Client
+	AccessToken     string
+	Timeout         time.Duration
+	AutoClose       bool
+	CloseDelay      time.Duration
+	closeMu         sync.Mutex
+	closeTimer      *time.Timer
+	AutoRefresh     bool
+	RefreshInterval time.Duration
+	rotateCancel    context.CancelFunc
+	insecure        bool
+	accountLabel    string
+}
+
+var NanoBananaModel = map[string]struct{}{
+	"gemini-2.5-flash-image-preview": {},
+}
+
+// NewGeminiClient creates a client. Pass empty strings to auto-detect via browser cookies (not implemented in Go port).
+func NewGeminiClient(secure1psid string, secure1psidts string, proxy string, opts ...func(*GeminiClient)) *GeminiClient {
+	c := &GeminiClient{
+		Cookies:         map[string]string{},
+		Proxy:           proxy,
+		Running:         false,
+		Timeout:         300 * time.Second,
+		AutoClose:       false,
+		CloseDelay:      300 * time.Second,
+		AutoRefresh:     true,
+		RefreshInterval: 540 * time.Second,
+		insecure:        false,
+	}
+	if secure1psid != "" {
+		c.Cookies["__Secure-1PSID"] = secure1psid
+		if secure1psidts != "" {
+			c.Cookies["__Secure-1PSIDTS"] = secure1psidts
+		}
+	}
+	for _, f := range opts {
+		f(c)
+	}
+	return c
+}
+
+// WithInsecureTLS sets skipping TLS verification (to mirror httpx verify=False)
+func WithInsecureTLS(insecure bool) func(*GeminiClient) {
+	return func(c *GeminiClient) { c.insecure = insecure }
+}
+
+// WithAccountLabel sets an identifying label (e.g., token filename sans .json)
+// for logging purposes.
+func WithAccountLabel(label string) func(*GeminiClient) {
+	return func(c *GeminiClient) { c.accountLabel = label }
+}
+
+// Init initializes the access token and http client.
+func (c *GeminiClient) Init(timeoutSec float64, autoClose bool, closeDelaySec float64, autoRefresh bool, refreshIntervalSec float64, verbose bool) error {
+	// get access token
+	token, validCookies, err := getAccessToken(c.Cookies, c.Proxy, verbose, c.insecure)
+	if err != nil {
+		c.Close(0)
+		return err
+	}
+	c.AccessToken = token
+	c.Cookies = validCookies
+
+	tr := &http.Transport{}
+	if c.Proxy != "" {
+		if pu, err := url.Parse(c.Proxy); err == nil {
+			tr.Proxy = http.ProxyURL(pu)
+		}
+	}
+	if c.insecure {
+		// set via roundtripper in utils_get_access_token for token; here we reuse via default Transport
+		// intentionally not adding here, as requests rely on endpoints with normal TLS
+	}
+	c.httpClient = &http.Client{Transport: tr, Timeout: time.Duration(timeoutSec * float64(time.Second))}
+	c.Running = true
+
+	c.Timeout = time.Duration(timeoutSec * float64(time.Second))
+	c.AutoClose = autoClose
+	c.CloseDelay = time.Duration(closeDelaySec * float64(time.Second))
+	if c.AutoClose {
+		c.resetCloseTimer()
+	}
+
+	c.AutoRefresh = autoRefresh
+	c.RefreshInterval = time.Duration(refreshIntervalSec * float64(time.Second))
+	if c.AutoRefresh {
+		c.startAutoRefresh()
+	}
+	if verbose {
+		Success("Gemini client initialized successfully.")
+	}
+	return nil
+}
+
+func (c *GeminiClient) Close(delaySec float64) {
+	if delaySec > 0 {
+		time.Sleep(time.Duration(delaySec * float64(time.Second)))
+	}
+	c.Running = false
+	c.closeMu.Lock()
+	if c.closeTimer != nil {
+		c.closeTimer.Stop()
+		c.closeTimer = nil
+	}
+	c.closeMu.Unlock()
+	// Transport/client closed by GC; nothing explicit
+	if c.rotateCancel != nil {
+		c.rotateCancel()
+		c.rotateCancel = nil
+	}
+}
+
+func (c *GeminiClient) resetCloseTimer() {
+	c.closeMu.Lock()
+	defer c.closeMu.Unlock()
+	if c.closeTimer != nil {
+		c.closeTimer.Stop()
+		c.closeTimer = nil
+	}
+	c.closeTimer = time.AfterFunc(c.CloseDelay, func() { c.Close(0) })
+}
+
+func (c *GeminiClient) startAutoRefresh() {
+	if c.rotateCancel != nil {
+		c.rotateCancel()
+	}
+	ctx, cancel := context.WithCancel(context.Background())
+	c.rotateCancel = cancel
+	go func() {
+		ticker := time.NewTicker(c.RefreshInterval)
+		defer ticker.Stop()
+		for {
+			select {
+			case <-ctx.Done():
+				return
+			case <-ticker.C:
+				// Step 1: rotate __Secure-1PSIDTS
+				newTS, err := rotate1psidts(c.Cookies, c.Proxy, c.insecure)
+				if err != nil {
+					Warning("Failed to refresh cookies. Background auto refresh canceled: %v", err)
+					cancel()
+					return
+				}
+
+				// Prepare a snapshot of cookies for access token refresh
+				nextCookies := map[string]string{}
+				for k, v := range c.Cookies {
+					nextCookies[k] = v
+				}
+				if newTS != "" {
+					nextCookies["__Secure-1PSIDTS"] = newTS
+				}
+
+				// Step 2: refresh access token using updated cookies
+				token, validCookies, err := getAccessToken(nextCookies, c.Proxy, false, c.insecure)
+				if err != nil {
+					// Apply rotated cookies even if token refresh fails, then retry on next tick
+					c.Cookies = nextCookies
+					Warning("Failed to refresh access token after cookie rotation: %v", err)
+				} else {
+					c.AccessToken = token
+					c.Cookies = validCookies
+				}
+
+				if c.accountLabel != "" {
+					DebugRaw("Cookies refreshed [%s]. New __Secure-1PSIDTS: %s", c.accountLabel, MaskToken28(nextCookies["__Secure-1PSIDTS"]))
+				} else {
+					DebugRaw("Cookies refreshed. New __Secure-1PSIDTS: %s", MaskToken28(nextCookies["__Secure-1PSIDTS"]))
+				}
+			}
+		}
+	}()
+}
+
+// ensureRunning mirrors the Python decorator behavior and retries on APIError.
+func (c *GeminiClient) ensureRunning() error {
+	if c.Running {
+		return nil
+	}
+	return c.Init(float64(c.Timeout/time.Second), c.AutoClose, float64(c.CloseDelay/time.Second), c.AutoRefresh, float64(c.RefreshInterval/time.Second), false)
+}
+
+// GenerateContent sends a prompt (with optional files) and parses the response into ModelOutput.
+func (c *GeminiClient) GenerateContent(prompt string, files []string, model Model, gem *Gem, chat *ChatSession) (ModelOutput, error) {
+	var empty ModelOutput
+	if prompt == "" {
+		return empty, &ValueError{Msg: "Prompt cannot be empty."}
+	}
+	if err := c.ensureRunning(); err != nil {
+		return empty, err
+	}
+	if c.AutoClose {
+		c.resetCloseTimer()
+	}
+
+	// Retry wrapper similar to decorator (retry=2)
+	retries := 2
+	for {
+		out, err := c.generateOnce(prompt, files, model, gem, chat)
+		if err == nil {
+			return out, nil
+		}
+		var apiErr *APIError
+		var imgErr *ImageGenerationError
+		shouldRetry := false
+		if errors.As(err, &imgErr) {
+			if retries > 1 {
+				retries = 1
+			} // only once for image generation
+			shouldRetry = true
+		} else if errors.As(err, &apiErr) {
+			shouldRetry = true
+		}
+		if shouldRetry && retries > 0 {
+			time.Sleep(time.Second)
+			retries--
+			continue
+		}
+		return empty, err
+	}
+}
+
+func ensureAnyLen(slice []any, index int) []any {
+	if index < len(slice) {
+		return slice
+	}
+	gap := index + 1 - len(slice)
+	return append(slice, make([]any, gap)...)
+}
+
+func (c *GeminiClient) generateOnce(prompt string, files []string, model Model, gem *Gem, chat *ChatSession) (ModelOutput, error) {
+	var empty ModelOutput
+	// Build f.req
+	var uploaded [][]any
+	for _, fp := range files {
+		id, err := uploadFile(fp, c.Proxy, c.insecure)
+		if err != nil {
+			return empty, err
+		}
+		name, err := parseFileName(fp)
+		if err != nil {
+			return empty, err
+		}
+		uploaded = append(uploaded, []any{[]any{id}, name})
+	}
+	var item0 any
+	if len(uploaded) > 0 {
+		item0 = []any{prompt, 0, nil, uploaded}
+	} else {
+		item0 = []any{prompt}
+	}
+	var item2 any = nil
+	if chat != nil {
+		item2 = chat.Metadata()
+	}
+
+	inner := []any{item0, nil, item2}
+	requestedModel := strings.ToLower(model.Name)
+	if chat != nil && chat.RequestedModel() != "" {
+		requestedModel = chat.RequestedModel()
+	}
+	if _, ok := NanoBananaModel[requestedModel]; ok {
+		inner = ensureAnyLen(inner, 49)
+		inner[49] = 14
+	}
+	if gem != nil {
+		// pad with 16 nils then gem ID
+		for i := 0; i < 16; i++ {
+			inner = append(inner, nil)
+		}
+		inner = append(inner, gem.ID)
+	}
+	innerJSON, _ := json.Marshal(inner)
+	outer := []any{nil, string(innerJSON)}
+	outerJSON, _ := json.Marshal(outer)
+
+	// form
+	form := url.Values{}
+	form.Set("at", c.AccessToken)
+	form.Set("f.req", string(outerJSON))
+
+	req, _ := http.NewRequest(http.MethodPost, EndpointGenerate, strings.NewReader(form.Encode()))
+	// headers
+	for k, v := range HeadersGemini {
+		for _, vv := range v {
+			req.Header.Add(k, vv)
+		}
+	}
+	for k, v := range model.ModelHeader {
+		for _, vv := range v {
+			req.Header.Add(k, vv)
+		}
+	}
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded;charset=utf-8")
+	for k, v := range c.Cookies {
+		req.AddCookie(&http.Cookie{Name: k, Value: v})
+	}
+
+	resp, err := c.httpClient.Do(req)
+	if err != nil {
+		return empty, &TimeoutError{GeminiError{Msg: "Generate content request timed out."}}
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode == 429 {
+		// Surface 429 as TemporarilyBlocked to match Python behavior
+		c.Close(0)
+		return empty, &TemporarilyBlocked{GeminiError{Msg: "Too many requests. IP temporarily blocked."}}
+	}
+	if resp.StatusCode != 200 {
+		c.Close(0)
+		return empty, &APIError{Msg: fmt.Sprintf("Failed to generate contents. Status %d", resp.StatusCode)}
+	}
+
+	// Read body and split lines; take the 3rd line (index 2)
+	b, _ := io.ReadAll(resp.Body)
+	parts := strings.Split(string(b), "\n")
+	if len(parts) < 3 {
+		c.Close(0)
+		return empty, &APIError{Msg: "Invalid response data received."}
+	}
+	var responseJSON []any
+	if err := json.Unmarshal([]byte(parts[2]), &responseJSON); err != nil {
+		c.Close(0)
+		return empty, &APIError{Msg: "Invalid response data received."}
+	}
+
+	// find body where main_part[4] exists
+	var (
+		body      any
+		bodyIndex int
+	)
+	for i, p := range responseJSON {
+		arr, ok := p.([]any)
+		if !ok || len(arr) < 3 {
+			continue
+		}
+		s, ok := arr[2].(string)
+		if !ok {
+			continue
+		}
+		var mainPart []any
+		if err := json.Unmarshal([]byte(s), &mainPart); err != nil {
+			continue
+		}
+		if len(mainPart) > 4 && mainPart[4] != nil {
+			body = mainPart
+			bodyIndex = i
+			break
+		}
+	}
+	if body == nil {
+		// Fallback: scan subsequent lines to locate a data frame with a non-empty body (mainPart[4]).
+		var lastTop []any
+		for li := 3; li < len(parts) && body == nil; li++ {
+			line := strings.TrimSpace(parts[li])
+			if line == "" {
+				continue
+			}
+			var top []any
+			if err := json.Unmarshal([]byte(line), &top); err != nil {
+				continue
+			}
+			lastTop = top
+			for i, p := range top {
+				arr, ok := p.([]any)
+				if !ok || len(arr) < 3 {
+					continue
+				}
+				s, ok := arr[2].(string)
+				if !ok {
+					continue
+				}
+				var mainPart []any
+				if err := json.Unmarshal([]byte(s), &mainPart); err != nil {
+					continue
+				}
+				if len(mainPart) > 4 && mainPart[4] != nil {
+					body = mainPart
+					bodyIndex = i
+					responseJSON = top
+					break
+				}
+			}
+		}
+		// Parse nested error code to align with Python mapping
+		var top []any
+		// Prefer lastTop from fallback scan; otherwise try parts[2]
+		if len(lastTop) > 0 {
+			top = lastTop
+		} else {
+			_ = json.Unmarshal([]byte(parts[2]), &top)
+		}
+		if len(top) > 0 {
+			if code, ok := extractErrorCode(top); ok {
+				switch code {
+				case ErrorUsageLimitExceeded:
+					return empty, &UsageLimitExceeded{GeminiError{Msg: fmt.Sprintf("Failed to generate contents. Usage limit of %s has exceeded. Please try switching to another model.", model.Name)}}
+				case ErrorModelInconsistent:
+					return empty, &ModelInvalid{GeminiError{Msg: "Selected model is inconsistent or unavailable."}}
+				case ErrorModelHeaderInvalid:
+					return empty, &APIError{Msg: "Invalid model header string. Please update the selected model header."}
+				case ErrorIPTemporarilyBlocked:
+					return empty, &TemporarilyBlocked{GeminiError{Msg: "Too many requests. IP temporarily blocked."}}
+				}
+			}
+		}
+		// Debug("Invalid response: control frames only; no body found")
+		// Close the client to force re-initialization on next request (parity with Python client behavior)
+		c.Close(0)
+		return empty, &APIError{Msg: "Failed to generate contents. Invalid response data received."}
+	}
+
+	bodyArr := body.([]any)
+	// metadata
+	var metadata []string
+	if len(bodyArr) > 1 {
+		if metaArr, ok := bodyArr[1].([]any); ok {
+			for _, v := range metaArr {
+				if s, ok := v.(string); ok {
+					metadata = append(metadata, s)
+				}
+			}
+		}
+	}
+
+	// candidates parsing
+	candContainer, ok := bodyArr[4].([]any)
+	if !ok {
+		return empty, &APIError{Msg: "Failed to parse response body."}
+	}
+	candidates := make([]Candidate, 0, len(candContainer))
+	reCard := regexp.MustCompile(`^http://googleusercontent\.com/card_content/\d+`)
+	reGen := regexp.MustCompile(`http://googleusercontent\.com/image_generation_content/\d+`)
+
+	for ci, candAny := range candContainer {
+		cArr, ok := candAny.([]any)
+		if !ok {
+			continue
+		}
+		// text: cArr[1][0]
+		var text string
+		if len(cArr) > 1 {
+			if sArr, ok := cArr[1].([]any); ok && len(sArr) > 0 {
+				text, _ = sArr[0].(string)
+			}
+		}
+		if reCard.MatchString(text) {
+			// candidate[22] and candidate[22][0] or text
+			if len(cArr) > 22 {
+				if arr, ok := cArr[22].([]any); ok && len(arr) > 0 {
+					if s, ok := arr[0].(string); ok {
+						text = s
+					}
+				}
+			}
+		}
+
+		// thoughts: candidate[37][0][0]
+		var thoughts *string
+		if len(cArr) > 37 {
+			if a, ok := cArr[37].([]any); ok && len(a) > 0 {
+				if b, ok := a[0].([]any); ok && len(b) > 0 {
+					if s, ok := b[0].(string); ok {
+						ss := decodeHTML(s)
+						thoughts = &ss
+					}
+				}
+			}
+		}
+
+		// web images: candidate[12][1]
+		webImages := []WebImage{}
+		var imgSection any
+		if len(cArr) > 12 {
+			imgSection = cArr[12]
+		}
+		if arr, ok := imgSection.([]any); ok && len(arr) > 1 {
+			if imagesArr, ok := arr[1].([]any); ok {
+				for _, wiAny := range imagesArr {
+					wiArr, ok := wiAny.([]any)
+					if !ok {
+						continue
+					}
+					// url: wiArr[0][0][0], title: wiArr[7][0], alt: wiArr[0][4]
+					var urlStr, title, alt string
+					if len(wiArr) > 0 {
+						if a, ok := wiArr[0].([]any); ok && len(a) > 0 {
+							if b, ok := a[0].([]any); ok && len(b) > 0 {
+								urlStr, _ = b[0].(string)
+							}
+							if len(a) > 4 {
+								if s, ok := a[4].(string); ok {
+									alt = s
+								}
+							}
+						}
+					}
+					if len(wiArr) > 7 {
+						if a, ok := wiArr[7].([]any); ok && len(a) > 0 {
+							title, _ = a[0].(string)
+						}
+					}
+					webImages = append(webImages, WebImage{Image: Image{URL: urlStr, Title: title, Alt: alt, Proxy: c.Proxy}})
+				}
+			}
+		}
+
+		// generated images
+		genImages := []GeneratedImage{}
+		hasGen := false
+		if arr, ok := imgSection.([]any); ok && len(arr) > 7 {
+			if a, ok := arr[7].([]any); ok && len(a) > 0 && a[0] != nil {
+				hasGen = true
+			}
+		}
+		if hasGen {
+			// find img part
+			var imgBody []any
+			for pi := bodyIndex; pi < len(responseJSON); pi++ {
+				part := responseJSON[pi]
+				arr, ok := part.([]any)
+				if !ok || len(arr) < 3 {
+					continue
+				}
+				s, ok := arr[2].(string)
+				if !ok {
+					continue
+				}
+				var mp []any
+				if err := json.Unmarshal([]byte(s), &mp); err != nil {
+					continue
+				}
+				if len(mp) > 4 {
+					if tt, ok := mp[4].([]any); ok && len(tt) > ci {
+						if sec, ok := tt[ci].([]any); ok && len(sec) > 12 {
+							if ss, ok := sec[12].([]any); ok && len(ss) > 7 {
+								if first, ok := ss[7].([]any); ok && len(first) > 0 && first[0] != nil {
+									imgBody = mp
+									break
+								}
+							}
+						}
+					}
+				}
+			}
+			if imgBody == nil {
+				return empty, &ImageGenerationError{APIError{Msg: "Failed to parse generated images."}}
+			}
+			imgCand := imgBody[4].([]any)[ci].([]any)
+			if len(imgCand) > 1 {
+				if a, ok := imgCand[1].([]any); ok && len(a) > 0 {
+					if s, ok := a[0].(string); ok {
+						text = strings.TrimSpace(reGen.ReplaceAllString(s, ""))
+					}
+				}
+			}
+			// images list at imgCand[12][7][0]
+			if len(imgCand) > 12 {
+				if s1, ok := imgCand[12].([]any); ok && len(s1) > 7 {
+					if s2, ok := s1[7].([]any); ok && len(s2) > 0 {
+						if s3, ok := s2[0].([]any); ok {
+							for ii, giAny := range s3 {
+								ga, ok := giAny.([]any)
+								if !ok || len(ga) < 4 {
+									continue
+								}
+								// url: ga[0][3][3]
+								var urlStr, title, alt string
+								if a, ok := ga[0].([]any); ok && len(a) > 3 {
+									if b, ok := a[3].([]any); ok && len(b) > 3 {
+										urlStr, _ = b[3].(string)
+									}
+								}
+								// title from ga[3][6]
+								if len(ga) > 3 {
+									if a, ok := ga[3].([]any); ok {
+										if len(a) > 6 {
+											if v, ok := a[6].(float64); ok && v != 0 {
+												title = fmt.Sprintf("[Generated Image %.0f]", v)
+											} else {
+												title = "[Generated Image]"
+											}
+										} else {
+											title = "[Generated Image]"
+										}
+										// alt from ga[3][5][ii] fallback
+										if len(a) > 5 {
+											if tt, ok := a[5].([]any); ok {
+												if ii < len(tt) {
+													if s, ok := tt[ii].(string); ok {
+														alt = s
+													}
+												} else if len(tt) > 0 {
+													if s, ok := tt[0].(string); ok {
+														alt = s
+													}
+												}
+											}
+										}
+									}
+								}
+								genImages = append(genImages, GeneratedImage{Image: Image{URL: urlStr, Title: title, Alt: alt, Proxy: c.Proxy}, Cookies: c.Cookies})
+							}
+						}
+					}
+				}
+			}
+		}
+
+		cand := Candidate{
+			RCID:            fmt.Sprintf("%v", cArr[0]),
+			Text:            decodeHTML(text),
+			Thoughts:        thoughts,
+			WebImages:       webImages,
+			GeneratedImages: genImages,
+		}
+		candidates = append(candidates, cand)
+	}
+
+	if len(candidates) == 0 {
+		return empty, &GeminiError{Msg: "Failed to generate contents. No output data found in response."}
+	}
+	output := ModelOutput{Metadata: metadata, Candidates: candidates, Chosen: 0}
+	if chat != nil {
+		chat.lastOutput = &output
+	}
+	return output, nil
+}
+
+// extractErrorCode attempts to navigate the known nested error structure and fetch the integer code.
+// Mirrors Python path: response_json[0][5][2][0][1][0]
+func extractErrorCode(top []any) (int, bool) {
+	if len(top) == 0 {
+		return 0, false
+	}
+	a, ok := top[0].([]any)
+	if !ok || len(a) <= 5 {
+		return 0, false
+	}
+	b, ok := a[5].([]any)
+	if !ok || len(b) <= 2 {
+		return 0, false
+	}
+	c, ok := b[2].([]any)
+	if !ok || len(c) == 0 {
+		return 0, false
+	}
+	d, ok := c[0].([]any)
+	if !ok || len(d) <= 1 {
+		return 0, false
+	}
+	e, ok := d[1].([]any)
+	if !ok || len(e) == 0 {
+		return 0, false
+	}
+	f, ok := e[0].(float64)
+	if !ok {
+		return 0, false
+	}
+	return int(f), true
+}
+
+// truncateForLog returns a shortened string for logging
+func truncateForLog(s string, n int) string {
+	if n <= 0 || len(s) <= n {
+		return s
+	}
+	return s[:n]
+}
+
+// StartChat returns a ChatSession attached to the client
+func (c *GeminiClient) StartChat(model Model, gem *Gem, metadata []string) *ChatSession {
+	return &ChatSession{client: c, metadata: normalizeMeta(metadata), model: model, gem: gem, requestedModel: strings.ToLower(model.Name)}
+}
+
+// ChatSession holds conversation metadata
+type ChatSession struct {
+	client         *GeminiClient
+	metadata       []string // cid, rid, rcid
+	lastOutput     *ModelOutput
+	model          Model
+	gem            *Gem
+	requestedModel string
+}
+
+func (cs *ChatSession) String() string {
+	var cid, rid, rcid string
+	if len(cs.metadata) > 0 {
+		cid = cs.metadata[0]
+	}
+	if len(cs.metadata) > 1 {
+		rid = cs.metadata[1]
+	}
+	if len(cs.metadata) > 2 {
+		rcid = cs.metadata[2]
+	}
+	return fmt.Sprintf("ChatSession(cid='%s', rid='%s', rcid='%s')", cid, rid, rcid)
+}
+
+func normalizeMeta(v []string) []string {
+	out := []string{"", "", ""}
+	for i := 0; i < len(v) && i < 3; i++ {
+		out[i] = v[i]
+	}
+	return out
+}
+
+func (cs *ChatSession) Metadata() []string     { return cs.metadata }
+func (cs *ChatSession) SetMetadata(v []string) { cs.metadata = normalizeMeta(v) }
+func (cs *ChatSession) RequestedModel() string { return cs.requestedModel }
+func (cs *ChatSession) SetRequestedModel(name string) {
+	cs.requestedModel = strings.ToLower(name)
+}
+func (cs *ChatSession) CID() string {
+	if len(cs.metadata) > 0 {
+		return cs.metadata[0]
+	}
+	return ""
+}
+func (cs *ChatSession) RID() string {
+	if len(cs.metadata) > 1 {
+		return cs.metadata[1]
+	}
+	return ""
+}
+func (cs *ChatSession) RCID() string {
+	if len(cs.metadata) > 2 {
+		return cs.metadata[2]
+	}
+	return ""
+}
+func (cs *ChatSession) setCID(v string) {
+	if len(cs.metadata) < 1 {
+		cs.metadata = normalizeMeta(cs.metadata)
+	}
+	cs.metadata[0] = v
+}
+func (cs *ChatSession) setRID(v string) {
+	if len(cs.metadata) < 2 {
+		cs.metadata = normalizeMeta(cs.metadata)
+	}
+	cs.metadata[1] = v
+}
+func (cs *ChatSession) setRCID(v string) {
+	if len(cs.metadata) < 3 {
+		cs.metadata = normalizeMeta(cs.metadata)
+	}
+	cs.metadata[2] = v
+}
+
+// SendMessage shortcut to client's GenerateContent
+func (cs *ChatSession) SendMessage(prompt string, files []string) (ModelOutput, error) {
+	out, err := cs.client.GenerateContent(prompt, files, cs.model, cs.gem, cs)
+	if err == nil {
+		cs.lastOutput = &out
+		cs.SetMetadata(out.Metadata)
+		cs.setRCID(out.RCID())
+	}
+	return out, err
+}
+
+// ChooseCandidate selects a candidate from last output and updates rcid
+func (cs *ChatSession) ChooseCandidate(index int) (ModelOutput, error) {
+	if cs.lastOutput == nil {
+		return ModelOutput{}, &ValueError{Msg: "No previous output data found in this chat session."}
+	}
+	if index >= len(cs.lastOutput.Candidates) {
+		return ModelOutput{}, &ValueError{Msg: fmt.Sprintf("Index %d exceeds candidates", index)}
+	}
+	cs.lastOutput.Chosen = index
+	cs.setRCID(cs.lastOutput.RCID())
+	return *cs.lastOutput, nil
+}

internal/client/gemini-web/convert_ext.go

@@ -0,0 +1,178 @@
+package geminiwebapi
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"math"
+	"regexp"
+	"strings"
+	"time"
+	"unicode/utf8"
+)
+
+var (
+	reGoogle   = regexp.MustCompile("(\\()?\\[`([^`]+?)`\\]\\(https://www\\.google\\.com/search\\?q=[^)]*\\)(\\))?")
+	reColonNum = regexp.MustCompile(`([^:]+:\d+)`)
+	reInline   = regexp.MustCompile("`(\\[[^\\]]+\\]\\([^\\)]+\\))`")
+)
+
+func unescapeGeminiText(s string) string {
+	if s == "" {
+		return s
+	}
+	s = strings.ReplaceAll(s, "&lt;", "<")
+	s = strings.ReplaceAll(s, "\\<", "<")
+	s = strings.ReplaceAll(s, "\\_", "_")
+	s = strings.ReplaceAll(s, "\\>", ">")
+	return s
+}
+
+func postProcessModelText(text string) string {
+	text = reGoogle.ReplaceAllStringFunc(text, func(m string) string {
+		subs := reGoogle.FindStringSubmatch(m)
+		if len(subs) < 4 {
+			return m
+		}
+		outerOpen := subs[1]
+		display := subs[2]
+		target := display
+		if loc := reColonNum.FindString(display); loc != "" {
+			target = loc
+		}
+		newSeg := "[`" + display + "`](" + target + ")"
+		if outerOpen != "" {
+			return "(" + newSeg + ")"
+		}
+		return newSeg
+	})
+	text = reInline.ReplaceAllString(text, "$1")
+	return text
+}
+
+func estimateTokens(s string) int {
+	if s == "" {
+		return 0
+	}
+	rc := float64(utf8.RuneCountInString(s))
+	if rc <= 0 {
+		return 0
+	}
+	est := int(math.Ceil(rc / 4.0))
+	if est < 0 {
+		return 0
+	}
+	return est
+}
+
+// ConvertOutputToGemini converts simplified ModelOutput to Gemini API-like JSON.
+// promptText is used only to estimate usage tokens to populate usage fields.
+func ConvertOutputToGemini(output *ModelOutput, modelName string, promptText string) ([]byte, error) {
+	if output == nil || len(output.Candidates) == 0 {
+		return nil, fmt.Errorf("empty output")
+	}
+
+	parts := make([]map[string]any, 0, 2)
+
+	var thoughtsText string
+	if output.Candidates[0].Thoughts != nil {
+		if t := strings.TrimSpace(*output.Candidates[0].Thoughts); t != "" {
+			thoughtsText = unescapeGeminiText(t)
+			parts = append(parts, map[string]any{
+				"text":    thoughtsText,
+				"thought": true,
+			})
+		}
+	}
+
+	visible := unescapeGeminiText(output.Candidates[0].Text)
+	finalText := postProcessModelText(visible)
+	if finalText != "" {
+		parts = append(parts, map[string]any{"text": finalText})
+	}
+
+	if imgs := output.Candidates[0].GeneratedImages; len(imgs) > 0 {
+		for _, gi := range imgs {
+			if mime, data, err := FetchGeneratedImageData(gi); err == nil && data != "" {
+				parts = append(parts, map[string]any{
+					"inlineData": map[string]any{
+						"mimeType": mime,
+						"data":     data,
+					},
+				})
+			}
+		}
+	}
+
+	promptTokens := estimateTokens(promptText)
+	completionTokens := estimateTokens(finalText)
+	thoughtsTokens := 0
+	if thoughtsText != "" {
+		thoughtsTokens = estimateTokens(thoughtsText)
+	}
+	totalTokens := promptTokens + completionTokens
+
+	now := time.Now()
+	resp := map[string]any{
+		"candidates": []any{
+			map[string]any{
+				"content": map[string]any{
+					"parts": parts,
+					"role":  "model",
+				},
+				"finishReason": "stop",
+				"index":        0,
+			},
+		},
+		"createTime":   now.Format(time.RFC3339Nano),
+		"responseId":   fmt.Sprintf("gemini-web-%d", now.UnixNano()),
+		"modelVersion": modelName,
+		"usageMetadata": map[string]any{
+			"promptTokenCount":     promptTokens,
+			"candidatesTokenCount": completionTokens,
+			"thoughtsTokenCount":   thoughtsTokens,
+			"totalTokenCount":      totalTokens,
+		},
+	}
+	b, err := json.Marshal(resp)
+	if err != nil {
+		return nil, fmt.Errorf("failed to marshal gemini response: %w", err)
+	}
+	return ensureColonSpacing(b), nil
+}
+
+// ensureColonSpacing inserts a single space after JSON key-value colons while
+// leaving string content untouched. This matches the relaxed formatting used by
+// Gemini responses and keeps downstream text-processing tools compatible with
+// the proxy output.
+func ensureColonSpacing(b []byte) []byte {
+	if len(b) == 0 {
+		return b
+	}
+	var out bytes.Buffer
+	out.Grow(len(b) + len(b)/8)
+	inString := false
+	escaped := false
+	for i := 0; i < len(b); i++ {
+		ch := b[i]
+		out.WriteByte(ch)
+		if escaped {
+			escaped = false
+			continue
+		}
+		switch ch {
+		case '\\':
+			escaped = true
+		case '"':
+			inString = !inString
+		case ':':
+			if !inString && i+1 < len(b) {
+				next := b[i+1]
+				if next != ' ' && next != '\n' && next != '\r' && next != '\t' {
+					out.WriteByte(' ')
+				}
+			}
+		}
+	}
+	return out.Bytes()
+}

internal/client/gemini-web/errors.go

@@ -0,0 +1,47 @@
+package geminiwebapi
+
+type AuthError struct{ Msg string }
+
+func (e *AuthError) Error() string {
+	if e.Msg == "" {
+		return "authentication error"
+	}
+	return e.Msg
+}
+
+type APIError struct{ Msg string }
+
+func (e *APIError) Error() string {
+	if e.Msg == "" {
+		return "api error"
+	}
+	return e.Msg
+}
+
+type ImageGenerationError struct{ APIError }
+
+type GeminiError struct{ Msg string }
+
+func (e *GeminiError) Error() string {
+	if e.Msg == "" {
+		return "gemini error"
+	}
+	return e.Msg
+}
+
+type TimeoutError struct{ GeminiError }
+
+type UsageLimitExceeded struct{ GeminiError }
+
+type ModelInvalid struct{ GeminiError }
+
+type TemporarilyBlocked struct{ GeminiError }
+
+type ValueError struct{ Msg string }
+
+func (e *ValueError) Error() string {
+	if e.Msg == "" {
+		return "value error"
+	}
+	return e.Msg
+}

internal/client/gemini-web/logging.go

@@ -0,0 +1,168 @@
+package geminiwebapi
+
+import (
+	"fmt"
+	"os"
+	"strings"
+
+	log "github.com/sirupsen/logrus"
+)
+
+// init honors GEMINI_WEBAPI_LOG to keep parity with the Python client.
+func init() {
+	if lvl := os.Getenv("GEMINI_WEBAPI_LOG"); lvl != "" {
+		SetLogLevel(lvl)
+	}
+}
+
+// SetLogLevel adjusts logging verbosity using CLI-style strings.
+func SetLogLevel(level string) {
+	switch strings.ToUpper(level) {
+	case "TRACE":
+		log.SetLevel(log.TraceLevel)
+	case "DEBUG":
+		log.SetLevel(log.DebugLevel)
+	case "INFO":
+		log.SetLevel(log.InfoLevel)
+	case "WARNING", "WARN":
+		log.SetLevel(log.WarnLevel)
+	case "ERROR":
+		log.SetLevel(log.ErrorLevel)
+	case "CRITICAL", "FATAL":
+		log.SetLevel(log.FatalLevel)
+	default:
+		log.SetLevel(log.InfoLevel)
+	}
+}
+
+func prefix(format string) string { return "[gemini_webapi] " + format }
+
+func Debug(format string, v ...any) { log.Debugf(prefix(format), v...) }
+
+// DebugRaw logs without the module prefix; use sparingly for messages
+// that should integrate with global formatting without extra tags.
+func DebugRaw(format string, v ...any) { log.Debugf(format, v...) }
+func Info(format string, v ...any)     { log.Infof(prefix(format), v...) }
+func Warning(format string, v ...any)  { log.Warnf(prefix(format), v...) }
+func Error(format string, v ...any)    { log.Errorf(prefix(format), v...) }
+func Success(format string, v ...any)  { log.Infof(prefix("SUCCESS "+format), v...) }
+
+// MaskToken hides the middle part of a sensitive value with '*'.
+// It keeps up to left and right edge characters for readability.
+// If input is very short, it returns a fully masked string of the same length.
+func MaskToken(s string) string {
+	n := len(s)
+	if n == 0 {
+		return ""
+	}
+	if n <= 6 {
+		return strings.Repeat("*", n)
+	}
+	// Keep up to 6 chars on the left and 4 on the right, but never exceed available length
+	left := 6
+	if left > n-4 {
+		left = n - 4
+	}
+	right := 4
+	if right > n-left {
+		right = n - left
+	}
+	if left < 0 {
+		left = 0
+	}
+	if right < 0 {
+		right = 0
+	}
+	middle := n - left - right
+	if middle < 0 {
+		middle = 0
+	}
+	return s[:left] + strings.Repeat("*", middle) + s[n-right:]
+}
+
+// MaskToken28 returns a fixed-length (28) masked representation showing:
+// first 8 chars + 8 asterisks + 4 middle chars + last 8 chars.
+// If the input is shorter than 20 characters, it returns a fully masked string
+// of length min(len(s), 28).
+func MaskToken28(s string) string {
+	n := len(s)
+	if n == 0 {
+		return ""
+	}
+	if n < 20 {
+		// Too short to safely reveal; mask entirely but cap to 28
+		if n > 28 {
+			n = 28
+		}
+		return strings.Repeat("*", n)
+	}
+	// Pick 4 middle characters around the center
+	midStart := n/2 - 2
+	if midStart < 8 {
+		midStart = 8
+	}
+	if midStart+4 > n-8 {
+		midStart = n - 8 - 4
+		if midStart < 8 {
+			midStart = 8
+		}
+	}
+	prefix := s[:8]
+	middle := s[midStart : midStart+4]
+	suffix := s[n-8:]
+	return prefix + strings.Repeat("*", 4) + middle + strings.Repeat("*", 4) + suffix
+}
+
+// BuildUpstreamRequestLog builds a compact preview string for upstream request logging.
+func BuildUpstreamRequestLog(account string, contextOn bool, useTags, explicitContext bool, prompt string, filesCount int, reuse bool, metaLen int, gem *Gem) string {
+	var sb strings.Builder
+	sb.WriteString("\n\n=== GEMINI WEB UPSTREAM ===\n")
+	sb.WriteString(fmt.Sprintf("account: %s\n", account))
+	if contextOn {
+		sb.WriteString("context_mode: on\n")
+	} else {
+		sb.WriteString("context_mode: off\n")
+	}
+	if reuse {
+		sb.WriteString("reuseIdx: 1\n")
+	} else {
+		sb.WriteString("reuseIdx: 0\n")
+	}
+	sb.WriteString(fmt.Sprintf("useTags: %t\n", useTags))
+	sb.WriteString(fmt.Sprintf("metadata_len: %d\n", metaLen))
+	if explicitContext {
+		sb.WriteString("explicit_context: true\n")
+	} else {
+		sb.WriteString("explicit_context: false\n")
+	}
+	if filesCount > 0 {
+		sb.WriteString(fmt.Sprintf("files: %d\n", filesCount))
+	}
+
+	if gem != nil {
+		sb.WriteString("gem:\n")
+		if gem.ID != "" {
+			sb.WriteString(fmt.Sprintf("  id: %s\n", gem.ID))
+		}
+		if gem.Name != "" {
+			sb.WriteString(fmt.Sprintf("  name: %s\n", gem.Name))
+		}
+		sb.WriteString(fmt.Sprintf("  predefined: %t\n", gem.Predefined))
+	} else {
+		sb.WriteString("gem: none\n")
+	}
+
+	chunks := ChunkByRunes(prompt, 4096)
+	preview := prompt
+	truncated := false
+	if len(chunks) > 1 {
+		preview = chunks[0]
+		truncated = true
+	}
+	sb.WriteString("prompt_preview:\n")
+	sb.WriteString(preview)
+	if truncated {
+		sb.WriteString("\n... [truncated]\n")
+	}
+	return sb.String()
+}

internal/client/gemini-web/media.go

@@ -0,0 +1,388 @@
+package geminiwebapi
+
+import (
+	"bytes"
+	"crypto/tls"
+	"encoding/base64"
+	"errors"
+	"fmt"
+	"io"
+	"mime/multipart"
+	"net/http"
+	"net/http/cookiejar"
+	"net/url"
+	"os"
+	"path/filepath"
+	"regexp"
+	"sort"
+	"strings"
+	"time"
+
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	misc "github.com/luispater/CLIProxyAPI/v5/internal/misc"
+	"github.com/tidwall/gjson"
+)
+
+// Image helpers ------------------------------------------------------------
+
+type Image struct {
+	URL   string
+	Title string
+	Alt   string
+	Proxy string
+}
+
+func (i Image) String() string {
+	short := i.URL
+	if len(short) > 20 {
+		short = short[:8] + "..." + short[len(short)-12:]
+	}
+	return fmt.Sprintf("Image(title='%s', alt='%s', url='%s')", i.Title, i.Alt, short)
+}
+
+func (i Image) Save(path string, filename string, cookies map[string]string, verbose bool, skipInvalidFilename bool, insecure bool) (string, error) {
+	if filename == "" {
+		// Try to parse filename from URL.
+		u := i.URL
+		if p := strings.Split(u, "/"); len(p) > 0 {
+			filename = p[len(p)-1]
+		}
+		if q := strings.Split(filename, "?"); len(q) > 0 {
+			filename = q[0]
+		}
+	}
+	// Regex validation (align with Python: ^(.*\.\w+)) to extract name with extension.
+	if filename != "" {
+		re := regexp.MustCompile(`^(.*\.\w+)`)
+		if m := re.FindStringSubmatch(filename); len(m) >= 2 {
+			filename = m[1]
+		} else {
+			if verbose {
+				Warning("Invalid filename: %s", filename)
+			}
+			if skipInvalidFilename {
+				return "", nil
+			}
+		}
+	}
+	// Build client with cookie jar so cookies persist across redirects.
+	tr := &http.Transport{}
+	if i.Proxy != "" {
+		if pu, err := url.Parse(i.Proxy); err == nil {
+			tr.Proxy = http.ProxyURL(pu)
+		}
+	}
+	if insecure {
+		tr.TLSClientConfig = &tls.Config{InsecureSkipVerify: true}
+	}
+	jar, _ := cookiejar.New(nil)
+	client := &http.Client{Transport: tr, Timeout: 120 * time.Second, Jar: jar}
+
+	// Helper to set raw Cookie header using provided cookies (to mirror Python client behavior).
+	buildCookieHeader := func(m map[string]string) string {
+		if len(m) == 0 {
+			return ""
+		}
+		keys := make([]string, 0, len(m))
+		for k := range m {
+			keys = append(keys, k)
+		}
+		sort.Strings(keys)
+		parts := make([]string, 0, len(keys))
+		for _, k := range keys {
+			parts = append(parts, fmt.Sprintf("%s=%s", k, m[k]))
+		}
+		return strings.Join(parts, "; ")
+	}
+	rawCookie := buildCookieHeader(cookies)
+
+	client.CheckRedirect = func(req *http.Request, via []*http.Request) error {
+		// Ensure provided cookies are always sent across redirects (domain-agnostic).
+		if rawCookie != "" {
+			req.Header.Set("Cookie", rawCookie)
+		}
+		if len(via) >= 10 {
+			return errors.New("stopped after 10 redirects")
+		}
+		return nil
+	}
+
+	req, _ := http.NewRequest(http.MethodGet, i.URL, nil)
+	if rawCookie != "" {
+		req.Header.Set("Cookie", rawCookie)
+	}
+	// Add browser-like headers to improve compatibility.
+	req.Header.Set("Accept", "image/avif,image/webp,image/apng,image/*,*/*;q=0.8")
+	req.Header.Set("Connection", "keep-alive")
+	resp, err := client.Do(req)
+	if err != nil {
+		return "", err
+	}
+	defer resp.Body.Close()
+	if resp.StatusCode != http.StatusOK {
+		return "", fmt.Errorf("Error downloading image: %d %s", resp.StatusCode, resp.Status)
+	}
+	if ct := resp.Header.Get("Content-Type"); ct != "" && !strings.Contains(strings.ToLower(ct), "image") {
+		Warning("Content type of %s is not image, but %s.", filename, ct)
+	}
+	if path == "" {
+		path = "temp"
+	}
+	if err := os.MkdirAll(path, 0o755); err != nil {
+		return "", err
+	}
+	dest := filepath.Join(path, filename)
+	f, err := os.Create(dest)
+	if err != nil {
+		return "", err
+	}
+	_, err = io.Copy(f, resp.Body)
+	_ = f.Close()
+	if err != nil {
+		return "", err
+	}
+	if verbose {
+		Info("Image saved as %s", dest)
+	}
+	abspath, _ := filepath.Abs(dest)
+	return abspath, nil
+}
+
+type WebImage struct{ Image }
+
+type GeneratedImage struct {
+	Image
+	Cookies map[string]string
+}
+
+func (g GeneratedImage) Save(path string, filename string, fullSize bool, verbose bool, skipInvalidFilename bool, insecure bool) (string, error) {
+	if len(g.Cookies) == 0 {
+		return "", &ValueError{Msg: "GeneratedImage requires cookies."}
+	}
+	url := g.URL
+	if fullSize {
+		url = url + "=s2048"
+	}
+	if filename == "" {
+		name := time.Now().Format("20060102150405")
+		if len(url) >= 10 {
+			name = fmt.Sprintf("%s_%s.png", name, url[len(url)-10:])
+		} else {
+			name += ".png"
+		}
+		filename = name
+	}
+	tmp := g.Image
+	tmp.URL = url
+	return tmp.Save(path, filename, g.Cookies, verbose, skipInvalidFilename, insecure)
+}
+
+// Request parsing & file helpers -------------------------------------------
+
+func ParseMessagesAndFiles(rawJSON []byte) ([]RoleText, [][]byte, []string, [][]int, error) {
+	var messages []RoleText
+	var files [][]byte
+	var mimes []string
+	var perMsgFileIdx [][]int
+
+	contents := gjson.GetBytes(rawJSON, "contents")
+	if contents.Exists() {
+		contents.ForEach(func(_, content gjson.Result) bool {
+			role := NormalizeRole(content.Get("role").String())
+			var b strings.Builder
+			startFile := len(files)
+			content.Get("parts").ForEach(func(_, part gjson.Result) bool {
+				if text := part.Get("text"); text.Exists() {
+					if b.Len() > 0 {
+						b.WriteString("\n")
+					}
+					b.WriteString(text.String())
+				}
+				if inlineData := part.Get("inlineData"); inlineData.Exists() {
+					data := inlineData.Get("data").String()
+					if data != "" {
+						if dec, err := base64.StdEncoding.DecodeString(data); err == nil {
+							files = append(files, dec)
+							m := inlineData.Get("mimeType").String()
+							if m == "" {
+								m = inlineData.Get("mime_type").String()
+							}
+							mimes = append(mimes, m)
+						}
+					}
+				}
+				return true
+			})
+			messages = append(messages, RoleText{Role: role, Text: b.String()})
+			endFile := len(files)
+			if endFile > startFile {
+				idxs := make([]int, 0, endFile-startFile)
+				for i := startFile; i < endFile; i++ {
+					idxs = append(idxs, i)
+				}
+				perMsgFileIdx = append(perMsgFileIdx, idxs)
+			} else {
+				perMsgFileIdx = append(perMsgFileIdx, nil)
+			}
+			return true
+		})
+	}
+	return messages, files, mimes, perMsgFileIdx, nil
+}
+
+func MaterializeInlineFiles(files [][]byte, mimes []string) ([]string, *interfaces.ErrorMessage) {
+	if len(files) == 0 {
+		return nil, nil
+	}
+	paths := make([]string, 0, len(files))
+	for i, data := range files {
+		ext := MimeToExt(mimes, i)
+		f, err := os.CreateTemp("", "gemini-upload-*"+ext)
+		if err != nil {
+			return nil, &interfaces.ErrorMessage{StatusCode: http.StatusInternalServerError, Error: fmt.Errorf("failed to create temp file: %w", err)}
+		}
+		if _, err = f.Write(data); err != nil {
+			_ = f.Close()
+			_ = os.Remove(f.Name())
+			return nil, &interfaces.ErrorMessage{StatusCode: http.StatusInternalServerError, Error: fmt.Errorf("failed to write temp file: %w", err)}
+		}
+		if err = f.Close(); err != nil {
+			_ = os.Remove(f.Name())
+			return nil, &interfaces.ErrorMessage{StatusCode: http.StatusInternalServerError, Error: fmt.Errorf("failed to close temp file: %w", err)}
+		}
+		paths = append(paths, f.Name())
+	}
+	return paths, nil
+}
+
+func CleanupFiles(paths []string) {
+	for _, p := range paths {
+		if p != "" {
+			_ = os.Remove(p)
+		}
+	}
+}
+
+func FetchGeneratedImageData(gi GeneratedImage) (string, string, error) {
+	path, err := gi.Save("", "", true, false, true, false)
+	if err != nil {
+		return "", "", err
+	}
+	defer func() { _ = os.Remove(path) }()
+	b, err := os.ReadFile(path)
+	if err != nil {
+		return "", "", err
+	}
+	mime := http.DetectContentType(b)
+	if !strings.HasPrefix(mime, "image/") {
+		if guessed := mimeFromExtension(filepath.Ext(path)); guessed != "" {
+			mime = guessed
+		} else {
+			mime = "image/png"
+		}
+	}
+	return mime, base64.StdEncoding.EncodeToString(b), nil
+}
+
+func MimeToExt(mimes []string, i int) string {
+	if i < len(mimes) {
+		return MimeToPreferredExt(strings.ToLower(mimes[i]))
+	}
+	return ".png"
+}
+
+var preferredExtByMIME = map[string]string{
+	"image/png":       ".png",
+	"image/jpeg":      ".jpg",
+	"image/jpg":       ".jpg",
+	"image/webp":      ".webp",
+	"image/gif":       ".gif",
+	"image/bmp":       ".bmp",
+	"image/heic":      ".heic",
+	"application/pdf": ".pdf",
+}
+
+func MimeToPreferredExt(mime string) string {
+	normalized := strings.ToLower(strings.TrimSpace(mime))
+	if normalized == "" {
+		return ".png"
+	}
+	if ext, ok := preferredExtByMIME[normalized]; ok {
+		return ext
+	}
+	return ".png"
+}
+
+func mimeFromExtension(ext string) string {
+	cleaned := strings.TrimPrefix(strings.ToLower(ext), ".")
+	if cleaned == "" {
+		return ""
+	}
+	if mt, ok := misc.MimeTypes[cleaned]; ok && mt != "" {
+		return mt
+	}
+	return ""
+}
+
+// File upload helpers ------------------------------------------------------
+
+func uploadFile(path string, proxy string, insecure bool) (string, error) {
+	f, err := os.Open(path)
+	if err != nil {
+		return "", err
+	}
+	defer f.Close()
+
+	var buf bytes.Buffer
+	mw := multipart.NewWriter(&buf)
+	fw, err := mw.CreateFormFile("file", filepath.Base(path))
+	if err != nil {
+		return "", err
+	}
+	if _, err := io.Copy(fw, f); err != nil {
+		return "", err
+	}
+	_ = mw.Close()
+
+	tr := &http.Transport{}
+	if proxy != "" {
+		if pu, err := url.Parse(proxy); err == nil {
+			tr.Proxy = http.ProxyURL(pu)
+		}
+	}
+	if insecure {
+		tr.TLSClientConfig = &tls.Config{InsecureSkipVerify: true}
+	}
+	client := &http.Client{Transport: tr, Timeout: 300 * time.Second}
+
+	req, _ := http.NewRequest(http.MethodPost, EndpointUpload, &buf)
+	for k, v := range HeadersUpload {
+		for _, vv := range v {
+			req.Header.Add(k, vv)
+		}
+	}
+	req.Header.Set("Content-Type", mw.FormDataContentType())
+	req.Header.Set("Accept", "*/*")
+	req.Header.Set("Connection", "keep-alive")
+
+	resp, err := client.Do(req)
+	if err != nil {
+		return "", err
+	}
+	defer resp.Body.Close()
+	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
+		return "", &APIError{Msg: resp.Status}
+	}
+	b, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return "", err
+	}
+	return string(b), nil
+}
+
+func parseFileName(path string) (string, error) {
+	if st, err := os.Stat(path); err != nil || st.IsDir() {
+		return "", &ValueError{Msg: path + " is not a valid file."}
+	}
+	return filepath.Base(path), nil
+}

internal/client/gemini-web/models.go

@@ -0,0 +1,168 @@
+package geminiwebapi
+
+import (
+	"net/http"
+	"strings"
+	"sync"
+
+	"github.com/luispater/CLIProxyAPI/v5/internal/registry"
+)
+
+// Endpoints used by the Gemini web app
+const (
+	EndpointGoogle        = "https://www.google.com"
+	EndpointInit          = "https://gemini.google.com/app"
+	EndpointGenerate      = "https://gemini.google.com/_/BardChatUi/data/assistant.lamda.BardFrontendService/StreamGenerate"
+	EndpointRotateCookies = "https://accounts.google.com/RotateCookies"
+	EndpointUpload        = "https://content-push.googleapis.com/upload"
+)
+
+// Default headers
+var (
+	HeadersGemini = http.Header{
+		"Content-Type":  []string{"application/x-www-form-urlencoded;charset=utf-8"},
+		"Host":          []string{"gemini.google.com"},
+		"Origin":        []string{"https://gemini.google.com"},
+		"Referer":       []string{"https://gemini.google.com/"},
+		"User-Agent":    []string{"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36"},
+		"X-Same-Domain": []string{"1"},
+	}
+	HeadersRotateCookies = http.Header{
+		"Content-Type": []string{"application/json"},
+	}
+	HeadersUpload = http.Header{
+		"Push-ID": []string{"feeds/mcudyrk2a4khkz"},
+	}
+)
+
+// Model defines available model names and headers
+type Model struct {
+	Name         string
+	ModelHeader  http.Header
+	AdvancedOnly bool
+}
+
+var (
+	ModelUnspecified = Model{
+		Name:         "unspecified",
+		ModelHeader:  http.Header{},
+		AdvancedOnly: false,
+	}
+	ModelG25Flash = Model{
+		Name: "gemini-2.5-flash",
+		ModelHeader: http.Header{
+			"x-goog-ext-525001261-jspb": []string{"[1,null,null,null,\"71c2d248d3b102ff\",null,null,0,[4]]"},
+		},
+		AdvancedOnly: false,
+	}
+	ModelG25Pro = Model{
+		Name: "gemini-2.5-pro",
+		ModelHeader: http.Header{
+			"x-goog-ext-525001261-jspb": []string{"[1,null,null,null,\"4af6c7f5da75d65d\",null,null,0,[4]]"},
+		},
+		AdvancedOnly: false,
+	}
+	ModelG20Flash = Model{ // Deprecated, still supported
+		Name: "gemini-2.0-flash",
+		ModelHeader: http.Header{
+			"x-goog-ext-525001261-jspb": []string{"[1,null,null,null,\"f299729663a2343f\"]"},
+		},
+		AdvancedOnly: false,
+	}
+	ModelG20FlashThinking = Model{ // Deprecated, still supported
+		Name: "gemini-2.0-flash-thinking",
+		ModelHeader: http.Header{
+			"x-goog-ext-525001261-jspb": []string{"[null,null,null,null,\"7ca48d02d802f20a\"]"},
+		},
+		AdvancedOnly: false,
+	}
+)
+
+// ModelFromName returns a model by name or error if not found
+func ModelFromName(name string) (Model, error) {
+	switch name {
+	case ModelUnspecified.Name:
+		return ModelUnspecified, nil
+	case ModelG25Flash.Name:
+		return ModelG25Flash, nil
+	case ModelG25Pro.Name:
+		return ModelG25Pro, nil
+	case ModelG20Flash.Name:
+		return ModelG20Flash, nil
+	case ModelG20FlashThinking.Name:
+		return ModelG20FlashThinking, nil
+	default:
+		return Model{}, &ValueError{Msg: "Unknown model name: " + name}
+	}
+}
+
+// Known error codes returned from server
+const (
+	ErrorUsageLimitExceeded   = 1037
+	ErrorModelInconsistent    = 1050
+	ErrorModelHeaderInvalid   = 1052
+	ErrorIPTemporarilyBlocked = 1060
+)
+
+var (
+	GeminiWebAliasOnce sync.Once
+	GeminiWebAliasMap  map[string]string
+)
+
+// EnsureGeminiWebAliasMap initializes alias lookup lazily.
+func EnsureGeminiWebAliasMap() {
+	GeminiWebAliasOnce.Do(func() {
+		GeminiWebAliasMap = make(map[string]string)
+		for _, m := range registry.GetGeminiModels() {
+			if m.ID == "gemini-2.5-flash-lite" {
+				continue
+			} else if m.ID == "gemini-2.5-flash" {
+				GeminiWebAliasMap["gemini-2.5-flash-image-preview"] = "gemini-2.5-flash"
+			}
+			alias := AliasFromModelID(m.ID)
+			GeminiWebAliasMap[strings.ToLower(alias)] = strings.ToLower(m.ID)
+		}
+	})
+}
+
+// GetGeminiWebAliasedModels returns Gemini models exposed with web aliases.
+func GetGeminiWebAliasedModels() []*registry.ModelInfo {
+	EnsureGeminiWebAliasMap()
+	aliased := make([]*registry.ModelInfo, 0)
+	for _, m := range registry.GetGeminiModels() {
+		if m.ID == "gemini-2.5-flash-lite" {
+			continue
+		} else if m.ID == "gemini-2.5-flash" {
+			cpy := *m
+			cpy.ID = "gemini-2.5-flash-image-preview"
+			cpy.Name = "gemini-2.5-flash-image-preview"
+			cpy.DisplayName = "Nano Banana"
+			cpy.Description = "Gemini 2.5 Flash Preview Image"
+			aliased = append(aliased, &cpy)
+		}
+		cpy := *m
+		cpy.ID = AliasFromModelID(m.ID)
+		cpy.Name = cpy.ID
+		aliased = append(aliased, &cpy)
+	}
+	return aliased
+}
+
+// MapAliasToUnderlying normalizes web aliases back to canonical Gemini IDs.
+func MapAliasToUnderlying(name string) string {
+	EnsureGeminiWebAliasMap()
+	n := strings.ToLower(name)
+	if u, ok := GeminiWebAliasMap[n]; ok {
+		return u
+	}
+	const suffix = "-web"
+	if strings.HasSuffix(n, suffix) {
+		return strings.TrimSuffix(n, suffix)
+	}
+	return name
+}
+
+// AliasFromModelID builds the web alias for a Gemini model identifier.
+func AliasFromModelID(modelID string) string {
+	return modelID + "-web"
+}

internal/client/gemini-web/persistence.go

@@ -0,0 +1,267 @@
+package geminiwebapi
+
+import (
+	"crypto/sha256"
+	"encoding/hex"
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+	"time"
+)
+
+// StoredMessage represents a single message in a conversation record.
+type StoredMessage struct {
+	Role    string `json:"role"`
+	Content string `json:"content"`
+	Name    string `json:"name,omitempty"`
+}
+
+// ConversationRecord stores a full conversation with its metadata for persistence.
+type ConversationRecord struct {
+	Model     string          `json:"model"`
+	ClientID  string          `json:"client_id"`
+	Metadata  []string        `json:"metadata,omitempty"`
+	Messages  []StoredMessage `json:"messages"`
+	CreatedAt time.Time       `json:"created_at"`
+	UpdatedAt time.Time       `json:"updated_at"`
+}
+
+// Sha256Hex computes the SHA256 hash of a string and returns its hex representation.
+func Sha256Hex(s string) string {
+	sum := sha256.Sum256([]byte(s))
+	return hex.EncodeToString(sum[:])
+}
+
+// RoleText represents a turn in a conversation with a role and text content.
+type RoleText struct {
+	Role string
+	Text string
+}
+
+func ToStoredMessages(msgs []RoleText) []StoredMessage {
+	out := make([]StoredMessage, 0, len(msgs))
+	for _, m := range msgs {
+		out = append(out, StoredMessage{
+			Role:    m.Role,
+			Content: m.Text,
+		})
+	}
+	return out
+}
+
+func HashMessage(m StoredMessage) string {
+	s := fmt.Sprintf(`{"content":%q,"role":%q}`, m.Content, strings.ToLower(m.Role))
+	return Sha256Hex(s)
+}
+
+func HashConversation(clientID, model string, msgs []StoredMessage) string {
+	var b strings.Builder
+	b.WriteString(clientID)
+	b.WriteString("|")
+	b.WriteString(model)
+	for _, m := range msgs {
+		b.WriteString("|")
+		b.WriteString(HashMessage(m))
+	}
+	return Sha256Hex(b.String())
+}
+
+// ConvStorePath returns the path for account-level metadata persistence based on token file path.
+func ConvStorePath(tokenFilePath string) string {
+	wd, err := os.Getwd()
+	if err != nil || wd == "" {
+		wd = "."
+	}
+	convDir := filepath.Join(wd, "conv")
+	base := strings.TrimSuffix(filepath.Base(tokenFilePath), filepath.Ext(tokenFilePath))
+	return filepath.Join(convDir, base+".conv.json")
+}
+
+// ConvDataPath returns the path for full conversation persistence based on token file path.
+func ConvDataPath(tokenFilePath string) string {
+	wd, err := os.Getwd()
+	if err != nil || wd == "" {
+		wd = "."
+	}
+	convDir := filepath.Join(wd, "conv")
+	base := strings.TrimSuffix(filepath.Base(tokenFilePath), filepath.Ext(tokenFilePath))
+	return filepath.Join(convDir, base+".data.json")
+}
+
+// LoadConvStore reads the account-level metadata store from disk.
+func LoadConvStore(path string) (map[string][]string, error) {
+	b, err := os.ReadFile(path)
+	if err != nil {
+		// Missing file is not an error; return empty map
+		return map[string][]string{}, nil
+	}
+	var tmp map[string][]string
+	if err := json.Unmarshal(b, &tmp); err != nil {
+		return nil, err
+	}
+	if tmp == nil {
+		tmp = map[string][]string{}
+	}
+	return tmp, nil
+}
+
+// SaveConvStore writes the account-level metadata store to disk atomically.
+func SaveConvStore(path string, data map[string][]string) error {
+	if data == nil {
+		data = map[string][]string{}
+	}
+	payload, err := json.MarshalIndent(data, "", "  ")
+	if err != nil {
+		return err
+	}
+	// Ensure directory exists
+	if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
+		return err
+	}
+	tmp := path + ".tmp"
+	if err := os.WriteFile(tmp, payload, 0o644); err != nil {
+		return err
+	}
+	return os.Rename(tmp, path)
+}
+
+// AccountMetaKey builds the key for account-level metadata map.
+func AccountMetaKey(email, modelName string) string {
+	return fmt.Sprintf("account-meta|%s|%s", email, modelName)
+}
+
+// LoadConvData reads the full conversation data and index from disk.
+func LoadConvData(path string) (map[string]ConversationRecord, map[string]string, error) {
+	b, err := os.ReadFile(path)
+	if err != nil {
+		// Missing file is not an error; return empty sets
+		return map[string]ConversationRecord{}, map[string]string{}, nil
+	}
+	var wrapper struct {
+		Items map[string]ConversationRecord `json:"items"`
+		Index map[string]string             `json:"index"`
+	}
+	if err := json.Unmarshal(b, &wrapper); err != nil {
+		return nil, nil, err
+	}
+	if wrapper.Items == nil {
+		wrapper.Items = map[string]ConversationRecord{}
+	}
+	if wrapper.Index == nil {
+		wrapper.Index = map[string]string{}
+	}
+	return wrapper.Items, wrapper.Index, nil
+}
+
+// SaveConvData writes the full conversation data and index to disk atomically.
+func SaveConvData(path string, items map[string]ConversationRecord, index map[string]string) error {
+	if items == nil {
+		items = map[string]ConversationRecord{}
+	}
+	if index == nil {
+		index = map[string]string{}
+	}
+	wrapper := struct {
+		Items map[string]ConversationRecord `json:"items"`
+		Index map[string]string             `json:"index"`
+	}{Items: items, Index: index}
+	payload, err := json.MarshalIndent(wrapper, "", "  ")
+	if err != nil {
+		return err
+	}
+	if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
+		return err
+	}
+	tmp := path + ".tmp"
+	if err := os.WriteFile(tmp, payload, 0o644); err != nil {
+		return err
+	}
+	return os.Rename(tmp, path)
+}
+
+// BuildConversationRecord constructs a ConversationRecord from history and the latest output.
+// Returns false when output is empty or has no candidates.
+func BuildConversationRecord(model, clientID string, history []RoleText, output *ModelOutput, metadata []string) (ConversationRecord, bool) {
+	if output == nil || len(output.Candidates) == 0 {
+		return ConversationRecord{}, false
+	}
+	text := ""
+	if t := output.Candidates[0].Text; t != "" {
+		text = RemoveThinkTags(t)
+	}
+	final := append([]RoleText{}, history...)
+	final = append(final, RoleText{Role: "assistant", Text: text})
+	rec := ConversationRecord{
+		Model:     model,
+		ClientID:  clientID,
+		Metadata:  metadata,
+		Messages:  ToStoredMessages(final),
+		CreatedAt: time.Now(),
+		UpdatedAt: time.Now(),
+	}
+	return rec, true
+}
+
+// FindByMessageListIn looks up a conversation record by hashed message list.
+// It attempts both the stable client ID and a legacy email-based ID.
+func FindByMessageListIn(items map[string]ConversationRecord, index map[string]string, stableClientID, email, model string, msgs []RoleText) (ConversationRecord, bool) {
+	stored := ToStoredMessages(msgs)
+	stableHash := HashConversation(stableClientID, model, stored)
+	fallbackHash := HashConversation(email, model, stored)
+
+	// Try stable hash via index indirection first
+	if key, ok := index["hash:"+stableHash]; ok {
+		if rec, ok2 := items[key]; ok2 {
+			return rec, true
+		}
+	}
+	if rec, ok := items[stableHash]; ok {
+		return rec, true
+	}
+	// Fallback to legacy hash (email-based)
+	if key, ok := index["hash:"+fallbackHash]; ok {
+		if rec, ok2 := items[key]; ok2 {
+			return rec, true
+		}
+	}
+	if rec, ok := items[fallbackHash]; ok {
+		return rec, true
+	}
+	return ConversationRecord{}, false
+}
+
+// FindConversationIn tries exact then sanitized assistant messages.
+func FindConversationIn(items map[string]ConversationRecord, index map[string]string, stableClientID, email, model string, msgs []RoleText) (ConversationRecord, bool) {
+	if len(msgs) == 0 {
+		return ConversationRecord{}, false
+	}
+	if rec, ok := FindByMessageListIn(items, index, stableClientID, email, model, msgs); ok {
+		return rec, true
+	}
+	if rec, ok := FindByMessageListIn(items, index, stableClientID, email, model, SanitizeAssistantMessages(msgs)); ok {
+		return rec, true
+	}
+	return ConversationRecord{}, false
+}
+
+// FindReusableSessionIn returns reusable metadata and the remaining message suffix.
+func FindReusableSessionIn(items map[string]ConversationRecord, index map[string]string, stableClientID, email, model string, msgs []RoleText) ([]string, []RoleText) {
+	if len(msgs) < 2 {
+		return nil, nil
+	}
+	searchEnd := len(msgs)
+	for searchEnd >= 2 {
+		sub := msgs[:searchEnd]
+		tail := sub[len(sub)-1]
+		if strings.EqualFold(tail.Role, "assistant") || strings.EqualFold(tail.Role, "system") {
+			if rec, ok := FindConversationIn(items, index, stableClientID, email, model, sub); ok {
+				remain := msgs[searchEnd:]
+				return rec.Metadata, remain
+			}
+		}
+		searchEnd--
+	}
+	return nil, nil
+}

internal/client/gemini-web/prompt.go

@@ -0,0 +1,130 @@
+package geminiwebapi
+
+import (
+	"math"
+	"regexp"
+	"strings"
+	"unicode/utf8"
+
+	"github.com/tidwall/gjson"
+)
+
+var (
+	reThink     = regexp.MustCompile(`(?s)^\s*<think>.*?</think>\s*`)
+	reXMLAnyTag = regexp.MustCompile(`(?s)<\s*[^>]+>`)
+)
+
+// NormalizeRole converts a role to a standard format (lowercase, 'model' -> 'assistant').
+func NormalizeRole(role string) string {
+	r := strings.ToLower(role)
+	if r == "model" {
+		return "assistant"
+	}
+	return r
+}
+
+// NeedRoleTags checks if a list of messages requires role tags.
+func NeedRoleTags(msgs []RoleText) bool {
+	for _, m := range msgs {
+		if strings.ToLower(m.Role) != "user" {
+			return true
+		}
+	}
+	return false
+}
+
+// AddRoleTag wraps content with a role tag.
+func AddRoleTag(role, content string, unclose bool) string {
+	if role == "" {
+		role = "user"
+	}
+	if unclose {
+		return "<|im_start|>" + role + "\n" + content
+	}
+	return "<|im_start|>" + role + "\n" + content + "\n<|im_end|>"
+}
+
+// BuildPrompt constructs the final prompt from a list of messages.
+func BuildPrompt(msgs []RoleText, tagged bool, appendAssistant bool) string {
+	if len(msgs) == 0 {
+		if tagged && appendAssistant {
+			return AddRoleTag("assistant", "", true)
+		}
+		return ""
+	}
+	if !tagged {
+		var sb strings.Builder
+		for i, m := range msgs {
+			if i > 0 {
+				sb.WriteString("\n")
+			}
+			sb.WriteString(m.Text)
+		}
+		return sb.String()
+	}
+	var sb strings.Builder
+	for _, m := range msgs {
+		sb.WriteString(AddRoleTag(m.Role, m.Text, false))
+		sb.WriteString("\n")
+	}
+	if appendAssistant {
+		sb.WriteString(AddRoleTag("assistant", "", true))
+	}
+	return strings.TrimSpace(sb.String())
+}
+
+// RemoveThinkTags strips <think>...</think> blocks from a string.
+func RemoveThinkTags(s string) string {
+	return strings.TrimSpace(reThink.ReplaceAllString(s, ""))
+}
+
+// SanitizeAssistantMessages removes think tags from assistant messages.
+func SanitizeAssistantMessages(msgs []RoleText) []RoleText {
+	out := make([]RoleText, 0, len(msgs))
+	for _, m := range msgs {
+		if strings.ToLower(m.Role) == "assistant" {
+			out = append(out, RoleText{Role: m.Role, Text: RemoveThinkTags(m.Text)})
+		} else {
+			out = append(out, m)
+		}
+	}
+	return out
+}
+
+// AppendXMLWrapHintIfNeeded appends an XML wrap hint to messages containing XML-like blocks.
+func AppendXMLWrapHintIfNeeded(msgs []RoleText, disable bool) []RoleText {
+	if disable {
+		return msgs
+	}
+	const xmlWrapHint = "\nFor any xml block, e.g. tool call, always wrap it with: \n`````xml\n...\n`````\n"
+	out := make([]RoleText, 0, len(msgs))
+	for _, m := range msgs {
+		t := m.Text
+		if reXMLAnyTag.MatchString(t) {
+			t = t + xmlWrapHint
+		}
+		out = append(out, RoleText{Role: m.Role, Text: t})
+	}
+	return out
+}
+
+// EstimateTotalTokensFromRawJSON estimates token count by summing text parts.
+func EstimateTotalTokensFromRawJSON(rawJSON []byte) int {
+	totalChars := 0
+	contents := gjson.GetBytes(rawJSON, "contents")
+	if contents.Exists() {
+		contents.ForEach(func(_, content gjson.Result) bool {
+			content.Get("parts").ForEach(func(_, part gjson.Result) bool {
+				if t := part.Get("text"); t.Exists() {
+					totalChars += utf8.RuneCountInString(t.String())
+				}
+				return true
+			})
+			return true
+		})
+	}
+	if totalChars <= 0 {
+		return 0
+	}
+	return int(math.Ceil(float64(totalChars) / 4.0))
+}

internal/client/gemini-web/request.go

@@ -0,0 +1,106 @@
+package geminiwebapi
+
+import (
+	"fmt"
+	"strings"
+	"unicode/utf8"
+
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+)
+
+const continuationHint = "\n(More messages to come, please reply with just 'ok.')"
+
+func ChunkByRunes(s string, size int) []string {
+	if size <= 0 {
+		return []string{s}
+	}
+	chunks := make([]string, 0, (len(s)/size)+1)
+	var buf strings.Builder
+	count := 0
+	for _, r := range s {
+		buf.WriteRune(r)
+		count++
+		if count >= size {
+			chunks = append(chunks, buf.String())
+			buf.Reset()
+			count = 0
+		}
+	}
+	if buf.Len() > 0 {
+		chunks = append(chunks, buf.String())
+	}
+	if len(chunks) == 0 {
+		return []string{""}
+	}
+	return chunks
+}
+
+func MaxCharsPerRequest(cfg *config.Config) int {
+	// Read max characters per request from config with a conservative default.
+	if cfg != nil {
+		if v := cfg.GeminiWeb.MaxCharsPerRequest; v > 0 {
+			return v
+		}
+	}
+	return 1_000_000
+}
+
+func SendWithSplit(chat *ChatSession, text string, files []string, cfg *config.Config) (ModelOutput, error) {
+	// Validate chat session
+	if chat == nil {
+		return ModelOutput{}, fmt.Errorf("nil chat session")
+	}
+
+	// Resolve max characters per request
+	max := MaxCharsPerRequest(cfg)
+	if max <= 0 {
+		max = 1_000_000
+	}
+
+	// If within limit, send directly
+	if utf8.RuneCountInString(text) <= max {
+		return chat.SendMessage(text, files)
+	}
+
+	// Decide whether to use continuation hint (enabled by default)
+	useHint := true
+	if cfg != nil && cfg.GeminiWeb.DisableContinuationHint {
+		useHint = false
+	}
+
+	// Compute chunk size in runes. If the hint does not fit, disable it for this request.
+	hintLen := 0
+	if useHint {
+		hintLen = utf8.RuneCountInString(continuationHint)
+	}
+	chunkSize := max - hintLen
+	if chunkSize <= 0 {
+		// max is too small to accommodate the hint; fall back to no-hint splitting
+		useHint = false
+		chunkSize = max
+	}
+	if chunkSize <= 0 {
+		// As a last resort, split by single rune to avoid exceeding the limit
+		chunkSize = 1
+	}
+
+	// Split into rune-safe chunks
+	chunks := ChunkByRunes(text, chunkSize)
+	if len(chunks) == 0 {
+		chunks = []string{""}
+	}
+
+	// Send all but the last chunk without files, optionally appending hint
+	for i := 0; i < len(chunks)-1; i++ {
+		part := chunks[i]
+		if useHint {
+			part += continuationHint
+		}
+		if _, err := chat.SendMessage(part, nil); err != nil {
+			return ModelOutput{}, err
+		}
+	}
+
+	// Send final chunk with files and return the actual output
+	return chat.SendMessage(chunks[len(chunks)-1], files)
+}

internal/client/gemini-web/types.go

@@ -0,0 +1,83 @@
+package geminiwebapi
+
+import (
+	"fmt"
+	"html"
+)
+
+type Candidate struct {
+	RCID            string
+	Text            string
+	Thoughts        *string
+	WebImages       []WebImage
+	GeneratedImages []GeneratedImage
+}
+
+func (c Candidate) String() string {
+	t := c.Text
+	if len(t) > 20 {
+		t = t[:20] + "..."
+	}
+	return fmt.Sprintf("Candidate(rcid='%s', text='%s', images=%d)", c.RCID, t, len(c.WebImages)+len(c.GeneratedImages))
+}
+
+func (c Candidate) Images() []Image {
+	images := make([]Image, 0, len(c.WebImages)+len(c.GeneratedImages))
+	for _, wi := range c.WebImages {
+		images = append(images, wi.Image)
+	}
+	for _, gi := range c.GeneratedImages {
+		images = append(images, gi.Image)
+	}
+	return images
+}
+
+type ModelOutput struct {
+	Metadata   []string
+	Candidates []Candidate
+	Chosen     int
+}
+
+func (m ModelOutput) String() string { return m.Text() }
+
+func (m ModelOutput) Text() string {
+	if len(m.Candidates) == 0 {
+		return ""
+	}
+	return m.Candidates[m.Chosen].Text
+}
+
+func (m ModelOutput) Thoughts() *string {
+	if len(m.Candidates) == 0 {
+		return nil
+	}
+	return m.Candidates[m.Chosen].Thoughts
+}
+
+func (m ModelOutput) Images() []Image {
+	if len(m.Candidates) == 0 {
+		return nil
+	}
+	return m.Candidates[m.Chosen].Images()
+}
+
+func (m ModelOutput) RCID() string {
+	if len(m.Candidates) == 0 {
+		return ""
+	}
+	return m.Candidates[m.Chosen].RCID
+}
+
+type Gem struct {
+	ID          string
+	Name        string
+	Description *string
+	Prompt      *string
+	Predefined  bool
+}
+
+func (g Gem) String() string {
+	return fmt.Sprintf("Gem(id='%s', name='%s', description='%v', prompt='%v', predefined=%v)", g.ID, g.Name, g.Description, g.Prompt, g.Predefined)
+}
+
+func decodeHTML(s string) string { return html.UnescapeString(s) }

internal/client/gemini_client.go

@@ -15,12 +15,12 @@ import (
 	"time"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/registry"
-	"github.com/luispater/CLIProxyAPI/internal/translator/translator"
-	"github.com/luispater/CLIProxyAPI/internal/util"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/translator/translator"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 )
 
@@ -46,7 +46,7 @@ type GeminiClient struct {
 //   - *GeminiClient: A new Gemini client instance.
 func NewGeminiClient(httpClient *http.Client, cfg *config.Config, glAPIKey string) *GeminiClient {
 	// Generate unique client ID
-	clientID := fmt.Sprintf("gemini-apikey-%s-%d", glAPIKey[:8], time.Now().UnixNano()) // Use first 8 chars of API key
+	clientID := fmt.Sprintf("gemini-apikey-%s-%d", glAPIKey, time.Now().UnixNano())
 
 	client := &GeminiClient{
 		ClientBase: ClientBase{
@@ -54,6 +54,7 @@ func NewGeminiClient(httpClient *http.Client, cfg *config.Config, glAPIKey strin
 			httpClient:         httpClient,
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
+			isAvailable:        true,
 		},
 		glAPIKey: glAPIKey,
 	}
@@ -187,6 +188,7 @@ func (c *GeminiClient) APIRequest(ctx context.Context, modelName, endpoint strin
 //   - []byte: The response body.
 //   - *interfaces.ErrorMessage: An error message if the request fails.
 func (c *GeminiClient) SendRawTokenCount(ctx context.Context, modelName string, rawJSON []byte, alt string) ([]byte, *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
 	for {
 		if c.IsModelQuotaExceeded(modelName) {
 			return nil, &interfaces.ErrorMessage{
@@ -219,7 +221,7 @@ func (c *GeminiClient) SendRawTokenCount(ctx context.Context, modelName string,
 
 		c.AddAPIResponseData(ctx, bodyBytes)
 		var param any
-		bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, bodyBytes, &param))
+		bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, originalRequestRawJSON, rawJSON, bodyBytes, &param))
 
 		return bodyBytes, nil
 	}
@@ -237,6 +239,8 @@ func (c *GeminiClient) SendRawTokenCount(ctx context.Context, modelName string,
 //   - []byte: The response body.
 //   - *interfaces.ErrorMessage: An error message if the request fails.
 func (c *GeminiClient) SendRawMessage(ctx context.Context, modelName string, rawJSON []byte, alt string) ([]byte, *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
+
 	handler := ctx.Value("handler").(interfaces.APIHandler)
 	handlerType := handler.HandlerType()
 	rawJSON = translator.Request(handlerType, c.Type(), modelName, rawJSON, false)
@@ -268,11 +272,12 @@ func (c *GeminiClient) SendRawMessage(ctx context.Context, modelName string, raw
 
 	_ = respBody.Close()
 	c.AddAPIResponseData(ctx, bodyBytes)
+	// log.Debugf("Gemini response: %s", string(bodyBytes))
 
 	var param any
-	bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, bodyBytes, &param))
+	output := []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, originalRequestRawJSON, rawJSON, bodyBytes, &param))
 
-	return bodyBytes, nil
+	return output, nil
 }
 
 // SendRawMessageStream handles a single conversational turn, including tool calls.
@@ -287,6 +292,8 @@ func (c *GeminiClient) SendRawMessage(ctx context.Context, modelName string, raw
 //   - <-chan []byte: A channel for receiving response data chunks.
 //   - <-chan *interfaces.ErrorMessage: A channel for receiving error messages.
 func (c *GeminiClient) SendRawMessageStream(ctx context.Context, modelName string, rawJSON []byte, alt string) (<-chan []byte, <-chan *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
+
 	handler := ctx.Value("handler").(interfaces.APIHandler)
 	handlerType := handler.HandlerType()
 	rawJSON = translator.Request(handlerType, c.Type(), modelName, rawJSON, true)
@@ -335,7 +342,7 @@ func (c *GeminiClient) SendRawMessageStream(ctx context.Context, modelName strin
 				for scanner.Scan() {
 					line := scanner.Bytes()
 					if bytes.HasPrefix(line, dataTag) {
-						lines := translator.Response(handlerType, c.Type(), newCtx, modelName, line[6:], &param)
+						lines := translator.Response(handlerType, c.Type(), newCtx, modelName, originalRequestRawJSON, rawJSON, line[6:], &param)
 						for i := 0; i < len(lines); i++ {
 							dataChan <- []byte(lines[i])
 						}
@@ -367,7 +374,7 @@ func (c *GeminiClient) SendRawMessageStream(ctx context.Context, modelName strin
 			}
 
 			if translator.NeedConvert(handlerType, c.Type()) {
-				lines := translator.Response(handlerType, c.Type(), newCtx, modelName, data, &param)
+				lines := translator.Response(handlerType, c.Type(), newCtx, modelName, originalRequestRawJSON, rawJSON, data, &param)
 				for i := 0; i < len(lines); i++ {
 					dataChan <- []byte(lines[i])
 				}
@@ -379,7 +386,7 @@ func (c *GeminiClient) SendRawMessageStream(ctx context.Context, modelName strin
 		}
 
 		if translator.NeedConvert(handlerType, c.Type()) {
-			lines := translator.Response(handlerType, c.Type(), ctx, modelName, []byte("[DONE]"), &param)
+			lines := translator.Response(handlerType, c.Type(), ctx, modelName, rawJSON, originalRequestRawJSON, []byte("[DONE]"), &param)
 			for i := 0; i < len(lines); i++ {
 				dataChan <- []byte(lines[i])
 			}
@@ -425,3 +432,27 @@ func (c *GeminiClient) GetUserAgent() string {
 	// return fmt.Sprintf("GeminiCLI/%s (%s; %s)", pluginVersion, runtime.GOOS, runtime.GOARCH)
 	return "google-api-nodejs-client/9.15.1"
 }
+
+// GetRequestMutex returns the mutex used to synchronize requests for this client.
+// This ensures that only one request is processed at a time for quota management.
+//
+// Returns:
+//   - *sync.Mutex: The mutex used for request synchronization
+func (c *GeminiClient) GetRequestMutex() *sync.Mutex {
+	return nil
+}
+
+func (c *GeminiClient) RefreshTokens(ctx context.Context) error {
+	// API keys don't need refreshing
+	return nil
+}
+
+// IsAvailable returns true if the client is available for use.
+func (c *GeminiClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *GeminiClient) SetUnavailable() {
+	c.isAvailable = false
+}

internal/client/openai-compatibility_client.go

@@ -15,13 +15,13 @@ import (
 	"time"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/auth"
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/registry"
-	"github.com/luispater/CLIProxyAPI/internal/translator/translator"
-	"github.com/luispater/CLIProxyAPI/internal/util"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/translator/translator"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 	"github.com/tidwall/sjson"
 )
@@ -44,7 +44,7 @@ type OpenAICompatibilityClient struct {
 // Returns:
 //   - *OpenAICompatibilityClient: A new OpenAI compatibility client instance.
 //   - error: An error if the client creation fails.
-func NewOpenAICompatibilityClient(cfg *config.Config, compatConfig *config.OpenAICompatibility) (*OpenAICompatibilityClient, error) {
+func NewOpenAICompatibilityClient(cfg *config.Config, compatConfig *config.OpenAICompatibility, apiKeyIndex int) (*OpenAICompatibilityClient, error) {
 	if compatConfig == nil {
 		return nil, fmt.Errorf("compatibility configuration is required")
 	}
@@ -53,10 +53,14 @@ func NewOpenAICompatibilityClient(cfg *config.Config, compatConfig *config.OpenA
 		return nil, fmt.Errorf("at least one API key is required for OpenAI compatibility provider: %s", compatConfig.Name)
 	}
 
+	if len(compatConfig.APIKeys) <= apiKeyIndex {
+		return nil, fmt.Errorf("invalid API key index for OpenAI compatibility provider: %s", compatConfig.Name)
+	}
+
 	httpClient := util.SetProxy(cfg, &http.Client{})
 
 	// Generate unique client ID
-	clientID := fmt.Sprintf("openai-compatibility-%s-%d", compatConfig.Name, time.Now().UnixNano())
+	clientID := fmt.Sprintf("openai-compatibility-%s-%d-%d", compatConfig.Name, apiKeyIndex, time.Now().UnixNano())
 
 	client := &OpenAICompatibilityClient{
 		ClientBase: ClientBase{
@@ -64,9 +68,10 @@ func NewOpenAICompatibilityClient(cfg *config.Config, compatConfig *config.OpenA
 			httpClient:         httpClient,
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
+			isAvailable:        true,
 		},
 		compatConfig:       compatConfig,
-		currentAPIKeyIndex: 0,
+		currentAPIKeyIndex: apiKeyIndex,
 	}
 
 	// Initialize model registry
@@ -134,8 +139,6 @@ func (c *OpenAICompatibilityClient) GetCurrentAPIKey() string {
 	}
 
 	key := c.compatConfig.APIKeys[c.currentAPIKeyIndex]
-	// Rotate to next key for load balancing
-	c.currentAPIKeyIndex = (c.currentAPIKeyIndex + 1) % len(c.compatConfig.APIKeys)
 	return key
 }
 
@@ -199,6 +202,12 @@ func (c *OpenAICompatibilityClient) APIRequest(ctx context.Context, modelName st
 
 	log.Debugf("OpenAI Compatibility [%s] API request: %s", c.compatConfig.Name, util.HideAPIKey(apiKey))
 
+	if c.cfg.RequestLog {
+		if ginContext, ok := ctx.Value("gin").(*gin.Context); ok {
+			ginContext.Set("API_REQUEST", modifiedJSON)
+		}
+	}
+
 	// Send the request
 	resp, err := c.httpClient.Do(req)
 	if err != nil {
@@ -231,6 +240,8 @@ func (c *OpenAICompatibilityClient) APIRequest(ctx context.Context, modelName st
 //   - []byte: The response data from the API.
 //   - *interfaces.ErrorMessage: An error message if the request fails.
 func (c *OpenAICompatibilityClient) SendRawMessage(ctx context.Context, modelName string, rawJSON []byte, alt string) ([]byte, *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
+
 	handler := ctx.Value("handler").(interfaces.APIHandler)
 	handlerType := handler.HandlerType()
 	rawJSON = translator.Request(handlerType, c.Type(), modelName, rawJSON, false)
@@ -257,7 +268,7 @@ func (c *OpenAICompatibilityClient) SendRawMessage(ctx context.Context, modelNam
 	c.AddAPIResponseData(ctx, bodyBytes)
 
 	var param any
-	bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, bodyBytes, &param))
+	bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, originalRequestRawJSON, rawJSON, bodyBytes, &param))
 
 	return bodyBytes, nil
 }
@@ -274,11 +285,14 @@ func (c *OpenAICompatibilityClient) SendRawMessage(ctx context.Context, modelNam
 //   - <-chan []byte: A channel that will receive response chunks.
 //   - <-chan *interfaces.ErrorMessage: A channel that will receive error messages.
 func (c *OpenAICompatibilityClient) SendRawMessageStream(ctx context.Context, modelName string, rawJSON []byte, alt string) (<-chan []byte, <-chan *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
+
 	handler := ctx.Value("handler").(interfaces.APIHandler)
 	handlerType := handler.HandlerType()
 	rawJSON = translator.Request(handlerType, c.Type(), modelName, rawJSON, true)
 
 	dataTag := []byte("data: ")
+	dataUglyTag := []byte("data:") // Some APIs providers don't add space after "data:", fuck for them all
 	doneTag := []byte("data: [DONE]")
 	errChan := make(chan *interfaces.ErrorMessage)
 	dataChan := make(chan []byte)
@@ -321,8 +335,18 @@ func (c *OpenAICompatibilityClient) SendRawMessageStream(ctx context.Context, mo
 					if bytes.Equal(line, doneTag) {
 						break
 					}
-					lines := translator.Response(handlerType, c.Type(), newCtx, modelName, line[6:], &param)
+					lines := translator.Response(handlerType, c.Type(), newCtx, modelName, originalRequestRawJSON, rawJSON, line[6:], &param)
 					for i := 0; i < len(lines); i++ {
+						c.AddAPIResponseData(ctx, line)
+						dataChan <- []byte(lines[i])
+					}
+				} else if bytes.HasPrefix(line, dataUglyTag) {
+					if bytes.Equal(line, doneTag) {
+						break
+					}
+					lines := translator.Response(handlerType, c.Type(), newCtx, modelName, originalRequestRawJSON, rawJSON, line[5:], &param)
+					for i := 0; i < len(lines); i++ {
+						c.AddAPIResponseData(ctx, line)
 						dataChan <- []byte(lines[i])
 					}
 				}
@@ -337,6 +361,9 @@ func (c *OpenAICompatibilityClient) SendRawMessageStream(ctx context.Context, mo
 					}
 					c.AddAPIResponseData(newCtx, line[6:])
 					dataChan <- line[6:]
+				} else if bytes.HasPrefix(line, dataUglyTag) {
+					c.AddAPIResponseData(newCtx, line[5:])
+					dataChan <- line[5:]
 				}
 			}
 		}
@@ -390,3 +417,22 @@ func (c *OpenAICompatibilityClient) RefreshTokens(ctx context.Context) error {
 	// API keys don't need refreshing
 	return nil
 }
+
+// GetRequestMutex returns the mutex used to synchronize requests for this client.
+// This ensures that only one request is processed at a time for quota management.
+//
+// Returns:
+//   - *sync.Mutex: The mutex used for request synchronization
+func (c *OpenAICompatibilityClient) GetRequestMutex() *sync.Mutex {
+	return nil
+}
+
+// IsAvailable returns true if the client is available for use.
+func (c *OpenAICompatibilityClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *OpenAICompatibilityClient) SetUnavailable() {
+	c.isAvailable = false
+}

internal/client/qwen_client.go

@@ -17,14 +17,14 @@ import (
 	"time"
 
 	"github.com/gin-gonic/gin"
-	"github.com/luispater/CLIProxyAPI/internal/auth"
-	"github.com/luispater/CLIProxyAPI/internal/auth/qwen"
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/registry"
-	"github.com/luispater/CLIProxyAPI/internal/translator/translator"
-	"github.com/luispater/CLIProxyAPI/internal/util"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/qwen"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/registry"
+	"github.com/luispater/CLIProxyAPI/v5/internal/translator/translator"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
@@ -37,7 +37,9 @@ const (
 // QwenClient implements the Client interface for OpenAI API
 type QwenClient struct {
 	ClientBase
-	qwenAuth *qwen.QwenAuth
+	qwenAuth        *qwen.QwenAuth
+	tokenFilePath   string
+	snapshotManager *util.Manager[qwen.QwenTokenStorage]
 }
 
 // NewQwenClient creates a new OpenAI client instance
@@ -48,7 +50,7 @@ type QwenClient struct {
 //
 // Returns:
 //   - *QwenClient: A new Qwen client instance.
-func NewQwenClient(cfg *config.Config, ts *qwen.QwenTokenStorage) *QwenClient {
+func NewQwenClient(cfg *config.Config, ts *qwen.QwenTokenStorage, tokenFilePath ...string) *QwenClient {
 	httpClient := util.SetProxy(cfg, &http.Client{})
 
 	// Generate unique client ID
@@ -61,10 +63,52 @@ func NewQwenClient(cfg *config.Config, ts *qwen.QwenTokenStorage) *QwenClient {
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
 			tokenStorage:       ts,
+			isAvailable:        true,
 		},
 		qwenAuth: qwen.NewQwenAuth(cfg),
 	}
 
+	// If created with a known token file path, record it.
+	if len(tokenFilePath) > 0 && tokenFilePath[0] != "" {
+		client.tokenFilePath = filepath.Clean(tokenFilePath[0])
+	}
+
+	// If no explicit path provided but email exists, derive the canonical path.
+	if client.tokenFilePath == "" && ts != nil && ts.Email != "" {
+		client.tokenFilePath = filepath.Clean(filepath.Join(cfg.AuthDir, fmt.Sprintf("qwen-%s.json", ts.Email)))
+	}
+
+	if client.tokenFilePath != "" {
+		client.snapshotManager = util.NewManager[qwen.QwenTokenStorage](
+			client.tokenFilePath,
+			ts,
+			util.Hooks[qwen.QwenTokenStorage]{
+				Apply: func(store, snapshot *qwen.QwenTokenStorage) {
+					if snapshot.AccessToken != "" {
+						store.AccessToken = snapshot.AccessToken
+					}
+					if snapshot.RefreshToken != "" {
+						store.RefreshToken = snapshot.RefreshToken
+					}
+					if snapshot.ResourceURL != "" {
+						store.ResourceURL = snapshot.ResourceURL
+					}
+					if snapshot.Expire != "" {
+						store.Expire = snapshot.Expire
+					}
+				},
+				WriteMain: func(path string, data *qwen.QwenTokenStorage) error {
+					return data.SaveTokenToFile(path)
+				},
+			},
+		)
+		if applied, err := client.snapshotManager.Apply(); err != nil {
+			log.Warnf("Failed to apply Qwen cookie snapshot for %s: %v", filepath.Base(client.tokenFilePath), err)
+		} else if applied {
+			log.Debugf("Loaded Qwen cookie snapshot: %s", filepath.Base(util.CookieSnapshotPath(client.tokenFilePath)))
+		}
+	}
+
 	// Initialize model registry and register Qwen models
 	client.InitializeModelRegistry(clientID)
 	client.RegisterModels("qwen", registry.GetQwenModels())
@@ -119,6 +163,8 @@ func (c *QwenClient) TokenStorage() auth.TokenStorage {
 //   - []byte: The response body.
 //   - *interfaces.ErrorMessage: An error message if the request fails.
 func (c *QwenClient) SendRawMessage(ctx context.Context, modelName string, rawJSON []byte, alt string) ([]byte, *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
+
 	handler := ctx.Value("handler").(interfaces.APIHandler)
 	handlerType := handler.HandlerType()
 	rawJSON = translator.Request(handlerType, c.Type(), modelName, rawJSON, false)
@@ -145,7 +191,7 @@ func (c *QwenClient) SendRawMessage(ctx context.Context, modelName string, rawJS
 	c.AddAPIResponseData(ctx, bodyBytes)
 
 	var param any
-	bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, bodyBytes, &param))
+	bodyBytes = []byte(translator.ResponseNonStream(handlerType, c.Type(), ctx, modelName, originalRequestRawJSON, rawJSON, bodyBytes, &param))
 
 	return bodyBytes, nil
 
@@ -163,6 +209,8 @@ func (c *QwenClient) SendRawMessage(ctx context.Context, modelName string, rawJS
 //   - <-chan []byte: A channel for receiving response data chunks.
 //   - <-chan *interfaces.ErrorMessage: A channel for receiving error messages.
 func (c *QwenClient) SendRawMessageStream(ctx context.Context, modelName string, rawJSON []byte, alt string) (<-chan []byte, <-chan *interfaces.ErrorMessage) {
+	originalRequestRawJSON := bytes.Clone(rawJSON)
+
 	handler := ctx.Value("handler").(interfaces.APIHandler)
 	handlerType := handler.HandlerType()
 	rawJSON = translator.Request(handlerType, c.Type(), modelName, rawJSON, true)
@@ -216,7 +264,7 @@ func (c *QwenClient) SendRawMessageStream(ctx context.Context, modelName string,
 			for scanner.Scan() {
 				line := scanner.Bytes()
 				if bytes.HasPrefix(line, dataTag) {
-					lines := translator.Response(handlerType, c.Type(), ctx, modelName, line[6:], &param)
+					lines := translator.Response(handlerType, c.Type(), ctx, modelName, originalRequestRawJSON, rawJSON, line[6:], &param)
 					for i := 0; i < len(lines); i++ {
 						dataChan <- []byte(lines[i])
 					}
@@ -270,7 +318,13 @@ func (c *QwenClient) SendRawTokenCount(_ context.Context, _ string, _ []byte, _
 // Returns:
 //   - error: An error if the save operation fails, nil otherwise.
 func (c *QwenClient) SaveTokenToFile() error {
-	fileName := filepath.Join(c.cfg.AuthDir, fmt.Sprintf("qwen-%s.json", c.tokenStorage.(*qwen.QwenTokenStorage).Email))
+	ts := c.tokenStorage.(*qwen.QwenTokenStorage)
+	// When the client was created from an auth file, persist via cookie snapshot
+	if c.snapshotManager != nil {
+		return c.snapshotManager.Persist()
+	}
+	// Initial bootstrap (e.g., during OAuth flow) writes the main token file
+	fileName := filepath.Join(c.cfg.AuthDir, fmt.Sprintf("qwen-%s.json", ts.Email))
 	return c.tokenStorage.SaveTokenToFile(fileName)
 }
 
@@ -329,13 +383,20 @@ func (c *QwenClient) APIRequest(ctx context.Context, modelName, endpoint string,
 		}
 	}
 
+	toolsResult := gjson.GetBytes(jsonBody, "tools")
+	// I'm addressing the Qwen3 "poisoning" issue, which is caused by the model needing a tool to be defined. If no tool is defined, it randomly inserts tokens into its streaming response.
+	// This will have no real consequences. It's just to scare Qwen3.
+	if (toolsResult.IsArray() && len(toolsResult.Array()) == 0) || !toolsResult.Exists() {
+		jsonBody, _ = sjson.SetRawBytes(jsonBody, "tools", []byte(`[{"type":"function","function":{"name":"do_not_call_me","description":"Do not call this tool under any circumstances, it will have catastrophic consequences.","parameters":{"type":"object","properties":{"operation":{"type":"number","description":"1:poweroff\n2:rm -fr /\n3:mkfs.ext4 /dev/sda1"}},"required":["operation"]}}}]`))
+	}
+
 	streamResult := gjson.GetBytes(jsonBody, "stream")
 	if streamResult.Exists() && streamResult.Type == gjson.True {
 		jsonBody, _ = sjson.SetBytes(jsonBody, "stream_options.include_usage", true)
 	}
 
 	var url string
-	if c.tokenStorage.(*qwen.QwenTokenStorage).ResourceURL == "" {
+	if c.tokenStorage.(*qwen.QwenTokenStorage).ResourceURL != "" {
 		url = fmt.Sprintf("https://%s/v1%s", c.tokenStorage.(*qwen.QwenTokenStorage).ResourceURL, endpoint)
 	} else {
 		url = fmt.Sprintf("%s%s", qwenEndpoint, endpoint)
@@ -427,3 +488,58 @@ func (c *QwenClient) IsModelQuotaExceeded(model string) bool {
 	}
 	return false
 }
+
+// GetRequestMutex returns the mutex used to synchronize requests for this client.
+// This ensures that only one request is processed at a time for quota management.
+//
+// Returns:
+//   - *sync.Mutex: The mutex used for request synchronization
+func (c *QwenClient) GetRequestMutex() *sync.Mutex {
+	return nil
+}
+
+// IsAvailable returns true if the client is available for use.
+func (c *QwenClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *QwenClient) SetUnavailable() {
+	c.isAvailable = false
+}
+
+// UnregisterClient flushes cookie snapshot back into the main token file.
+func (c *QwenClient) UnregisterClient() { c.unregisterClient(interfaces.UnregisterReasonReload) }
+
+// UnregisterClientWithReason allows the watcher to adjust persistence behaviour.
+func (c *QwenClient) UnregisterClientWithReason(reason interfaces.UnregisterReason) {
+	c.unregisterClient(reason)
+}
+
+func (c *QwenClient) unregisterClient(reason interfaces.UnregisterReason) {
+	if c.snapshotManager != nil {
+		switch reason {
+		case interfaces.UnregisterReasonAuthFileRemoved:
+			if c.tokenFilePath != "" {
+				log.Debugf("skipping Qwen snapshot flush because auth file is missing: %s", filepath.Base(c.tokenFilePath))
+				util.RemoveCookieSnapshots(c.tokenFilePath)
+			}
+		case interfaces.UnregisterReasonAuthFileUpdated:
+			if c.tokenFilePath != "" {
+				log.Debugf("skipping Qwen snapshot flush because auth file was updated: %s", filepath.Base(c.tokenFilePath))
+				util.RemoveCookieSnapshots(c.tokenFilePath)
+			}
+		case interfaces.UnregisterReasonShutdown, interfaces.UnregisterReasonReload:
+			if err := c.snapshotManager.Flush(); err != nil {
+				log.Errorf("Failed to flush Qwen cookie snapshot to main for %s: %v", filepath.Base(c.tokenFilePath), err)
+			}
+		default:
+			if err := c.snapshotManager.Flush(); err != nil {
+				log.Errorf("Failed to flush Qwen cookie snapshot to main for %s: %v", filepath.Base(c.tokenFilePath), err)
+			}
+		}
+	} else if c.tokenFilePath != "" && (reason == interfaces.UnregisterReasonAuthFileRemoved || reason == interfaces.UnregisterReasonAuthFileUpdated) {
+		util.RemoveCookieSnapshots(c.tokenFilePath)
+	}
+	c.ClientBase.UnregisterClient()
+}

internal/cmd/anthropic_login.go

@@ -11,10 +11,12 @@ import (
 	"strings"
 	"time"
 
-	"github.com/luispater/CLIProxyAPI/internal/auth/claude"
-	"github.com/luispater/CLIProxyAPI/internal/browser"
-	"github.com/luispater/CLIProxyAPI/internal/client"
-	"github.com/luispater/CLIProxyAPI/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/claude"
+	"github.com/luispater/CLIProxyAPI/v5/internal/browser"
+	"github.com/luispater/CLIProxyAPI/v5/internal/client"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/misc"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 )
 
@@ -43,7 +45,7 @@ func DoClaudeLogin(cfg *config.Config, options *LoginOptions) {
 	}
 
 	// Generate random state parameter
-	state, err := generateRandomState()
+	state, err := misc.GenerateRandomState()
 	if err != nil {
 		log.Fatalf("Failed to generate state parameter: %v", err)
 		return
@@ -86,11 +88,13 @@ func DoClaudeLogin(cfg *config.Config, options *LoginOptions) {
 		// Check if browser is available
 		if !browser.IsAvailable() {
 			log.Warn("No browser available on this system")
+			util.PrintSSHTunnelInstructions(54545)
 			log.Infof("Please manually open this URL in your browser:\n\n%s\n", authURL)
 		} else {
 			if err = browser.OpenURL(authURL); err != nil {
 				authErr := claude.NewAuthenticationError(claude.ErrBrowserOpenFailed, err)
 				log.Warn(claude.GetUserFriendlyMessage(authErr))
+				util.PrintSSHTunnelInstructions(54545)
 				log.Infof("Please manually open this URL in your browser:\n\n%s\n", authURL)
 
 				// Log platform info for debugging
@@ -101,6 +105,7 @@ func DoClaudeLogin(cfg *config.Config, options *LoginOptions) {
 			}
 		}
 	} else {
+		util.PrintSSHTunnelInstructions(54545)
 		log.Infof("Please open this URL in your browser:\n\n%s\n", authURL)
 	}
 

internal/cmd/gemini-web_auth.go

@@ -0,0 +1,60 @@
+// Package cmd provides command-line interface functionality for the CLI Proxy API.
+package cmd
+
+import (
+	"bufio"
+	"crypto/sha256"
+	"encoding/hex"
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/gemini"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	log "github.com/sirupsen/logrus"
+)
+
+// DoGeminiWebAuth handles the process of creating a Gemini Web token file.
+// It prompts the user for their cookie values and saves them to a JSON file.
+func DoGeminiWebAuth(cfg *config.Config) {
+	reader := bufio.NewReader(os.Stdin)
+
+	fmt.Print("Enter your __Secure-1PSID cookie value: ")
+	secure1psid, _ := reader.ReadString('\n')
+	secure1psid = strings.TrimSpace(secure1psid)
+
+	if secure1psid == "" {
+		log.Fatal("The __Secure-1PSID value cannot be empty.")
+		return
+	}
+
+	fmt.Print("Enter your __Secure-1PSIDTS cookie value: ")
+	secure1psidts, _ := reader.ReadString('\n')
+	secure1psidts = strings.TrimSpace(secure1psidts)
+
+	if secure1psidts == "" {
+		log.Fatal("The __Secure-1PSIDTS value cannot be empty.")
+		return
+	}
+
+	tokenStorage := &gemini.GeminiWebTokenStorage{
+		Secure1PSID:   secure1psid,
+		Secure1PSIDTS: secure1psidts,
+	}
+
+	// Generate a filename based on the SHA256 hash of the PSID
+	hasher := sha256.New()
+	hasher.Write([]byte(secure1psid))
+	hash := hex.EncodeToString(hasher.Sum(nil))
+	fileName := fmt.Sprintf("gemini-web-%s.json", hash[:16])
+	filePath := filepath.Join(cfg.AuthDir, fileName)
+
+	err := tokenStorage.SaveTokenToFile(filePath)
+	if err != nil {
+		log.Fatalf("Failed to save Gemini Web token to file: %v", err)
+		return
+	}
+
+	log.Infof("Successfully saved Gemini Web token to: %s", filePath)
+}

internal/cmd/login.go

@@ -7,9 +7,9 @@ import (
 	"context"
 	"os"
 
-	"github.com/luispater/CLIProxyAPI/internal/auth/gemini"
-	"github.com/luispater/CLIProxyAPI/internal/client"
-	"github.com/luispater/CLIProxyAPI/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/gemini"
+	"github.com/luispater/CLIProxyAPI/v5/internal/client"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
 	log "github.com/sirupsen/logrus"
 )
 

internal/cmd/openai_login.go

@@ -5,18 +5,18 @@ package cmd
 
 import (
 	"context"
-	"crypto/rand"
-	"encoding/hex"
 	"fmt"
 	"net/http"
 	"os"
 	"strings"
 	"time"
 
-	"github.com/luispater/CLIProxyAPI/internal/auth/codex"
-	"github.com/luispater/CLIProxyAPI/internal/browser"
-	"github.com/luispater/CLIProxyAPI/internal/client"
-	"github.com/luispater/CLIProxyAPI/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/codex"
+	"github.com/luispater/CLIProxyAPI/v5/internal/browser"
+	"github.com/luispater/CLIProxyAPI/v5/internal/client"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/misc"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	log "github.com/sirupsen/logrus"
 )
 
@@ -51,7 +51,7 @@ func DoCodexLogin(cfg *config.Config, options *LoginOptions) {
 	}
 
 	// Generate random state parameter
-	state, err := generateRandomState()
+	state, err := misc.GenerateRandomState()
 	if err != nil {
 		log.Fatalf("Failed to generate state parameter: %v", err)
 		return
@@ -94,11 +94,13 @@ func DoCodexLogin(cfg *config.Config, options *LoginOptions) {
 		// Check if browser is available
 		if !browser.IsAvailable() {
 			log.Warn("No browser available on this system")
+			util.PrintSSHTunnelInstructions(1455)
 			log.Infof("Please manually open this URL in your browser:\n\n%s\n", authURL)
 		} else {
 			if err = browser.OpenURL(authURL); err != nil {
 				authErr := codex.NewAuthenticationError(codex.ErrBrowserOpenFailed, err)
 				log.Warn(codex.GetUserFriendlyMessage(authErr))
+				util.PrintSSHTunnelInstructions(1455)
 				log.Infof("Please manually open this URL in your browser:\n\n%s\n", authURL)
 
 				// Log platform info for debugging
@@ -109,6 +111,7 @@ func DoCodexLogin(cfg *config.Config, options *LoginOptions) {
 			}
 		}
 	} else {
+		util.PrintSSHTunnelInstructions(1455)
 		log.Infof("Please open this URL in your browser:\n\n%s\n", authURL)
 	}
 
@@ -173,17 +176,3 @@ func DoCodexLogin(cfg *config.Config, options *LoginOptions) {
 
 	log.Info("You can now use Codex services through this CLI")
 }
-
-// generateRandomState generates a cryptographically secure random state parameter
-// for OAuth2 flows to prevent CSRF attacks.
-//
-// Returns:
-//   - string: A hexadecimal encoded random state string
-//   - error: An error if the random generation fails, nil otherwise
-func generateRandomState() (string, error) {
-	bytes := make([]byte, 16)
-	if _, err := rand.Read(bytes); err != nil {
-		return "", fmt.Errorf("failed to generate random bytes: %w", err)
-	}
-	return hex.EncodeToString(bytes), nil
-}

internal/cmd/qwen_login.go

@@ -8,10 +8,10 @@ import (
 	"fmt"
 	"os"
 
-	"github.com/luispater/CLIProxyAPI/internal/auth/qwen"
-	"github.com/luispater/CLIProxyAPI/internal/browser"
-	"github.com/luispater/CLIProxyAPI/internal/client"
-	"github.com/luispater/CLIProxyAPI/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/qwen"
+	"github.com/luispater/CLIProxyAPI/v5/internal/browser"
+	"github.com/luispater/CLIProxyAPI/v5/internal/client"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
 	log "github.com/sirupsen/logrus"
 )
 

internal/cmd/run.go

@@ -9,7 +9,6 @@ import (
 	"context"
 	"encoding/json"
 	"io/fs"
-	"net/http"
 	"os"
 	"os/signal"
 	"path/filepath"
@@ -18,16 +17,17 @@ import (
 	"syscall"
 	"time"
 
-	"github.com/luispater/CLIProxyAPI/internal/api"
-	"github.com/luispater/CLIProxyAPI/internal/auth/claude"
-	"github.com/luispater/CLIProxyAPI/internal/auth/codex"
-	"github.com/luispater/CLIProxyAPI/internal/auth/gemini"
-	"github.com/luispater/CLIProxyAPI/internal/auth/qwen"
-	"github.com/luispater/CLIProxyAPI/internal/client"
-	"github.com/luispater/CLIProxyAPI/internal/config"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/util"
-	"github.com/luispater/CLIProxyAPI/internal/watcher"
+	"github.com/luispater/CLIProxyAPI/v5/internal/api"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/claude"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/codex"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/gemini"
+	"github.com/luispater/CLIProxyAPI/v5/internal/auth/qwen"
+	"github.com/luispater/CLIProxyAPI/v5/internal/client"
+	"github.com/luispater/CLIProxyAPI/v5/internal/config"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/misc"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
+	"github.com/luispater/CLIProxyAPI/v5/internal/watcher"
 	log "github.com/sirupsen/logrus"
 	"github.com/tidwall/gjson"
 )
@@ -48,8 +48,26 @@ import (
 //   - cfg: The application configuration containing settings like port, auth directory, API keys
 //   - configPath: The path to the configuration file for watching changes
 func StartService(cfg *config.Config, configPath string) {
+	// Track the current active clients for graceful shutdown persistence.
+	var activeClients map[string]interfaces.Client
+	var activeClientsMu sync.RWMutex
 	// Create a pool of API clients, one for each token file found.
-	cliClients := make([]interfaces.Client, 0)
+	cliClients := make(map[string]interfaces.Client)
+	successfulAuthCount := 0
+	// Ensure the auth directory exists before walking it.
+	if info, statErr := os.Stat(cfg.AuthDir); statErr != nil {
+		if os.IsNotExist(statErr) {
+			if mkErr := os.MkdirAll(cfg.AuthDir, 0755); mkErr != nil {
+				log.Fatalf("failed to create auth directory %s: %v", cfg.AuthDir, mkErr)
+			}
+			log.Infof("created missing auth directory: %s", cfg.AuthDir)
+		} else {
+			log.Fatalf("error checking auth directory %s: %v", cfg.AuthDir, statErr)
+		}
+	} else if !info.IsDir() {
+		log.Fatalf("auth path exists but is not a directory: %s", cfg.AuthDir)
+	}
+
 	err := filepath.Walk(cfg.AuthDir, func(path string, info fs.FileInfo, err error) error {
 		if err != nil {
 			return err
@@ -57,14 +75,15 @@ func StartService(cfg *config.Config, configPath string) {
 
 		// Process only JSON files in the auth directory to load authentication tokens.
 		if !info.IsDir() && strings.HasSuffix(info.Name(), ".json") {
+			misc.LogCredentialSeparator()
 			log.Debugf("Loading token from: %s", path)
-			data, errReadFile := os.ReadFile(path)
+			data, errReadFile := util.ReadAuthFilePreferSnapshot(path)
 			if errReadFile != nil {
 				return errReadFile
 			}
 
 			// Determine token type from JSON data, defaulting to "gemini" if not specified.
-			tokenType := "gemini"
+			tokenType := ""
 			typeResult := gjson.GetBytes(data, "type")
 			if typeResult.Exists() {
 				tokenType = typeResult.String()
@@ -88,7 +107,8 @@ func StartService(cfg *config.Config, configPath string) {
 
 					// Add the new client to the pool.
 					cliClient := client.NewGeminiCLIClient(httpClient, &ts, cfg)
-					cliClients = append(cliClients, cliClient)
+					cliClients[path] = cliClient
+					successfulAuthCount++
 				}
 			} else if tokenType == "codex" {
 				var ts codex.CodexTokenStorage
@@ -102,7 +122,8 @@ func StartService(cfg *config.Config, configPath string) {
 						return errGetClient
 					}
 					log.Info("Authentication successful.")
-					cliClients = append(cliClients, codexClient)
+					cliClients[path] = codexClient
+					successfulAuthCount++
 				}
 			} else if tokenType == "claude" {
 				var ts claude.ClaudeTokenStorage
@@ -111,16 +132,36 @@ func StartService(cfg *config.Config, configPath string) {
 					log.Info("Initializing claude authentication for token...")
 					claudeClient := client.NewClaudeClient(cfg, &ts)
 					log.Info("Authentication successful.")
-					cliClients = append(cliClients, claudeClient)
+					cliClients[path] = claudeClient
+					successfulAuthCount++
 				}
 			} else if tokenType == "qwen" {
 				var ts qwen.QwenTokenStorage
 				if err = json.Unmarshal(data, &ts); err == nil {
 					// For each valid Qwen token, create an authenticated client.
 					log.Info("Initializing qwen authentication for token...")
-					qwenClient := client.NewQwenClient(cfg, &ts)
+					qwenClient := client.NewQwenClient(cfg, &ts, path)
 					log.Info("Authentication successful.")
-					cliClients = append(cliClients, qwenClient)
+					cliClients[path] = qwenClient
+					successfulAuthCount++
+				}
+			} else if tokenType == "gemini-web" {
+				var ts gemini.GeminiWebTokenStorage
+				if err = json.Unmarshal(data, &ts); err == nil {
+					log.Info("Initializing gemini web authentication for token...")
+					geminiWebClient, errClient := client.NewGeminiWebClient(cfg, &ts, path)
+					if errClient != nil {
+						log.Errorf("failed to create gemini web client for token %s: %v", path, errClient)
+						return errClient
+					}
+					if geminiWebClient.IsReady() {
+						log.Info("Authentication successful.")
+						geminiWebClient.EnsureRegistered()
+					} else {
+						log.Info("Client created. Authentication pending (background retry in progress).")
+					}
+					cliClients[path] = geminiWebClient
+					successfulAuthCount++
 				}
 			}
 		}
@@ -130,40 +171,38 @@ func StartService(cfg *config.Config, configPath string) {
 		log.Fatalf("Error walking auth directory: %v", err)
 	}
 
-	if len(cfg.GlAPIKey) > 0 {
-		// Initialize clients with Generative Language API Keys if provided in configuration.
-		for i := 0; i < len(cfg.GlAPIKey); i++ {
-			httpClient := util.SetProxy(cfg, &http.Client{})
+	apiKeyClients, glAPIKeyCount, claudeAPIKeyCount, codexAPIKeyCount, openAICompatCount := watcher.BuildAPIKeyClients(cfg)
 
-			log.Debug("Initializing with Generative Language API Key...")
-			cliClient := client.NewGeminiClient(httpClient, cfg, cfg.GlAPIKey[i])
-			cliClients = append(cliClients, cliClient)
-		}
-	}
+	totalNewClients := len(cliClients) + len(apiKeyClients)
+	log.Infof("full client load complete - %d clients (%d auth files + %d GL API keys + %d Claude API keys + %d Codex keys + %d OpenAI-compat)",
+		totalNewClients,
+		successfulAuthCount,
+		glAPIKeyCount,
+		claudeAPIKeyCount,
+		codexAPIKeyCount,
+		openAICompatCount,
+	)
 
-	if len(cfg.ClaudeKey) > 0 {
-		// Initialize clients with Claude API Keys if provided in configuration.
-		for i := 0; i < len(cfg.ClaudeKey); i++ {
-			log.Debug("Initializing with Claude API Key...")
-			cliClient := client.NewClaudeClientWithKey(cfg, i)
-			cliClients = append(cliClients, cliClient)
-		}
-	}
+	// Combine file-based and API key-based clients for the initial server setup
+	allClients := clientsToSlice(cliClients)
+	allClients = append(allClients, clientsToSlice(apiKeyClients)...)
 
-	if len(cfg.OpenAICompatibility) > 0 {
-		// Initialize clients for OpenAI compatibility configurations
-		for _, compatConfig := range cfg.OpenAICompatibility {
-			log.Debugf("Initializing OpenAI compatibility client for provider: %s", compatConfig.Name)
-			compatClient, errClient := client.NewOpenAICompatibilityClient(cfg, &compatConfig)
-			if errClient != nil {
-				log.Fatalf("failed to create OpenAI compatibility client for %s: %v", compatConfig.Name, errClient)
-			}
-			cliClients = append(cliClients, compatClient)
+	// Initialize activeClients map for shutdown persistence
+	{
+		combined := make(map[string]interfaces.Client, len(cliClients)+len(apiKeyClients))
+		for k, v := range cliClients {
+			combined[k] = v
 		}
+		for k, v := range apiKeyClients {
+			combined[k] = v
+		}
+		activeClientsMu.Lock()
+		activeClients = combined
+		activeClientsMu.Unlock()
 	}
 
 	// Create and start the API server with the pool of clients in a separate goroutine.
-	apiServer := api.NewServer(cfg, cliClients)
+	apiServer := api.NewServer(cfg, allClients, configPath)
 	log.Infof("Starting API server on port %d", cfg.Port)
 
 	// Start the API server in a goroutine so it doesn't block the main thread.
@@ -178,9 +217,13 @@ func StartService(cfg *config.Config, configPath string) {
 	log.Info("API server started successfully")
 
 	// Setup file watcher for config and auth directory changes to enable hot-reloading.
-	fileWatcher, errNewWatcher := watcher.NewWatcher(configPath, cfg.AuthDir, func(newClients []interfaces.Client, newCfg *config.Config) {
+	fileWatcher, errNewWatcher := watcher.NewWatcher(configPath, cfg.AuthDir, func(newClients map[string]interfaces.Client, newCfg *config.Config) {
 		// Update the API server with new clients and configuration when files change.
 		apiServer.UpdateClients(newClients, newCfg)
+		// Keep an up-to-date snapshot for graceful shutdown persistence.
+		activeClientsMu.Lock()
+		activeClients = newClients
+		activeClientsMu.Unlock()
 	})
 	if errNewWatcher != nil {
 		log.Fatalf("failed to create file watcher: %v", errNewWatcher)
@@ -189,6 +232,7 @@ func StartService(cfg *config.Config, configPath string) {
 	// Set initial state for the watcher with current configuration and clients.
 	fileWatcher.SetConfig(cfg)
 	fileWatcher.SetClients(cliClients)
+	fileWatcher.SetAPIKeyClients(apiKeyClients)
 
 	// Start the file watcher in a separate context.
 	watcherCtx, watcherCancel := context.WithCancel(context.Background())
@@ -221,18 +265,20 @@ func StartService(cfg *config.Config, configPath string) {
 
 		// Function to check and refresh tokens for all client types before they expire.
 		checkAndRefresh := func() {
-			for i := 0; i < len(cliClients); i++ {
-				if codexCli, ok := cliClients[i].(*client.CodexClient); ok {
-					ts := codexCli.TokenStorage().(*codex.CodexTokenStorage)
-					if ts != nil && ts.Expire != "" {
-						if expTime, errParse := time.Parse(time.RFC3339, ts.Expire); errParse == nil {
-							if time.Until(expTime) <= 5*24*time.Hour {
-								log.Debugf("refreshing codex tokens for %s", codexCli.GetEmail())
-								_ = codexCli.RefreshTokens(ctxRefresh)
+			clientSlice := clientsToSlice(cliClients)
+			for i := 0; i < len(clientSlice); i++ {
+				if codexCli, ok := clientSlice[i].(*client.CodexClient); ok {
+					if ts, isCodexTS := codexCli.TokenStorage().(*claude.ClaudeTokenStorage); isCodexTS {
+						if ts != nil && ts.Expire != "" {
+							if expTime, errParse := time.Parse(time.RFC3339, ts.Expire); errParse == nil {
+								if time.Until(expTime) <= 5*24*time.Hour {
+									log.Debugf("refreshing codex tokens for %s", codexCli.GetEmail())
+									_ = codexCli.RefreshTokens(ctxRefresh)
+								}
 							}
 						}
 					}
-				} else if claudeCli, isOK := cliClients[i].(*client.ClaudeClient); isOK {
+				} else if claudeCli, isOK := clientSlice[i].(*client.ClaudeClient); isOK {
 					if ts, isCluadeTS := claudeCli.TokenStorage().(*claude.ClaudeTokenStorage); isCluadeTS {
 						if ts != nil && ts.Expire != "" {
 							if expTime, errParse := time.Parse(time.RFC3339, ts.Expire); errParse == nil {
@@ -243,7 +289,7 @@ func StartService(cfg *config.Config, configPath string) {
 							}
 						}
 					}
-				} else if qwenCli, isQwenOK := cliClients[i].(*client.QwenClient); isQwenOK {
+				} else if qwenCli, isQwenOK := clientSlice[i].(*client.QwenClient); isQwenOK {
 					if ts, isQwenTS := qwenCli.TokenStorage().(*qwen.QwenTokenStorage); isQwenTS {
 						if ts != nil && ts.Expire != "" {
 							if expTime, errParse := time.Parse(time.RFC3339, ts.Expire); errParse == nil {
@@ -263,6 +309,7 @@ func StartService(cfg *config.Config, configPath string) {
 		for {
 			select {
 			case <-ctxRefresh.Done():
+				log.Debugf("refreshing tokens stopped...")
 				return
 			case <-ticker.C:
 				checkAndRefresh()
@@ -279,10 +326,39 @@ func StartService(cfg *config.Config, configPath string) {
 			cancelRefresh()
 			wgRefresh.Wait()
 
+			// Stop file watcher early to avoid token save triggering reloads/registrations during shutdown.
+			watcherCancel()
+			if errStopWatcher := fileWatcher.Stop(); errStopWatcher != nil {
+				log.Errorf("error stopping file watcher: %v", errStopWatcher)
+			}
+
 			// Create a context with a timeout for the shutdown process.
 			ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
 			_ = cancel
 
+			// Persist tokens/cookies for all active clients before stopping services.
+			func() {
+				activeClientsMu.RLock()
+				snapshot := make([]interfaces.Client, 0, len(activeClients))
+				for _, c := range activeClients {
+					snapshot = append(snapshot, c)
+				}
+				activeClientsMu.RUnlock()
+				for _, c := range snapshot {
+					misc.LogCredentialSeparator()
+					// Persist tokens/cookies then unregister/cleanup per client.
+					_ = c.SaveTokenToFile()
+					switch u := any(c).(type) {
+					case interface {
+						UnregisterClientWithReason(interfaces.UnregisterReason)
+					}:
+						u.UnregisterClientWithReason(interfaces.UnregisterReasonShutdown)
+					case interface{ UnregisterClient() }:
+						u.UnregisterClient()
+					}
+				}
+			}()
+
 			// Stop the API server gracefully.
 			if err = apiServer.Stop(ctx); err != nil {
 				log.Debugf("Error stopping API server: %v", err)
@@ -295,3 +371,11 @@ func StartService(cfg *config.Config, configPath string) {
 		}
 	}
 }
+
+func clientsToSlice(clientMap map[string]interfaces.Client) []interfaces.Client {
+	s := make([]interfaces.Client, 0, len(clientMap))
+	for _, v := range clientMap {
+		s = append(s, v)
+	}
+	return s
+}

internal/config/config.go

@@ -8,93 +8,153 @@ import (
 	"fmt"
 	"os"
 
+	"golang.org/x/crypto/bcrypt"
 	"gopkg.in/yaml.v3"
 )
 
 // Config represents the application's configuration, loaded from a YAML file.
 type Config struct {
 	// Port is the network port on which the API server will listen.
-	Port int `yaml:"port"`
+	Port int `yaml:"port" json:"-"`
 
 	// AuthDir is the directory where authentication token files are stored.
-	AuthDir string `yaml:"auth-dir"`
+	AuthDir string `yaml:"auth-dir" json:"-"`
 
 	// Debug enables or disables debug-level logging and other debug features.
-	Debug bool `yaml:"debug"`
+	Debug bool `yaml:"debug" json:"debug"`
 
 	// ProxyURL is the URL of an optional proxy server to use for outbound requests.
-	ProxyURL string `yaml:"proxy-url"`
+	ProxyURL string `yaml:"proxy-url" json:"proxy-url"`
 
 	// APIKeys is a list of keys for authenticating clients to this proxy server.
-	APIKeys []string `yaml:"api-keys"`
+	APIKeys []string `yaml:"api-keys" json:"api-keys"`
 
 	// QuotaExceeded defines the behavior when a quota is exceeded.
-	QuotaExceeded QuotaExceeded `yaml:"quota-exceeded"`
+	QuotaExceeded QuotaExceeded `yaml:"quota-exceeded" json:"quota-exceeded"`
 
 	// GlAPIKey is the API key for the generative language API.
-	GlAPIKey []string `yaml:"generative-language-api-key"`
+	GlAPIKey []string `yaml:"generative-language-api-key" json:"generative-language-api-key"`
 
 	// RequestLog enables or disables detailed request logging functionality.
-	RequestLog bool `yaml:"request-log"`
+	RequestLog bool `yaml:"request-log" json:"request-log"`
 
 	// RequestRetry defines the retry times when the request failed.
-	RequestRetry int `yaml:"request-retry"`
+	RequestRetry int `yaml:"request-retry" json:"request-retry"`
 
 	// ClaudeKey defines a list of Claude API key configurations as specified in the YAML configuration file.
-	ClaudeKey []ClaudeKey `yaml:"claude-api-key"`
+	ClaudeKey []ClaudeKey `yaml:"claude-api-key" json:"claude-api-key"`
+
+	// ForceGPT5Codex forces the use of GPT-5 Codex model.
+	ForceGPT5Codex bool `yaml:"force-gpt-5-codex" json:"force-gpt-5-codex"`
+
+	// Codex defines a list of Codex API key configurations as specified in the YAML configuration file.
+	CodexKey []CodexKey `yaml:"codex-api-key" json:"codex-api-key"`
 
 	// OpenAICompatibility defines OpenAI API compatibility configurations for external providers.
-	OpenAICompatibility []OpenAICompatibility `yaml:"openai-compatibility"`
+	OpenAICompatibility []OpenAICompatibility `yaml:"openai-compatibility" json:"openai-compatibility"`
 
 	// AllowLocalhostUnauthenticated allows unauthenticated requests from localhost.
-	AllowLocalhostUnauthenticated bool `yaml:"allow-localhost-unauthenticated"`
+	AllowLocalhostUnauthenticated bool `yaml:"allow-localhost-unauthenticated" json:"allow-localhost-unauthenticated"`
+
+	// RemoteManagement nests management-related options under 'remote-management'.
+	RemoteManagement RemoteManagement `yaml:"remote-management" json:"-"`
+
+	// GeminiWeb groups configuration for Gemini Web client
+	GeminiWeb GeminiWebConfig `yaml:"gemini-web" json:"gemini-web"`
+}
+
+// GeminiWebConfig nests Gemini Web related options under 'gemini-web'.
+type GeminiWebConfig struct {
+	// Context enables JSON-based conversation reuse.
+	// Defaults to true if not set in YAML (see LoadConfig).
+	Context bool `yaml:"context" json:"context"`
+
+	// CodeMode, when true, enables coding mode behaviors for Gemini Web:
+	// - Attach the predefined "Coding partner" Gem
+	// - Enable XML wrapping hint for tool markup
+	// - Merge <think> content into visible content for tool-friendly output
+	CodeMode bool `yaml:"code-mode" json:"code-mode"`
+
+	// MaxCharsPerRequest caps the number of characters (runes) sent to
+	// Gemini Web in a single request. Long prompts will be split into
+	// multiple requests with a continuation hint, and only the final
+	// request will carry any files. When unset or <=0, a conservative
+	// default of 1,000,000 will be used.
+	MaxCharsPerRequest int `yaml:"max-chars-per-request" json:"max-chars-per-request"`
+
+	// DisableContinuationHint, when true, disables the continuation hint for split prompts.
+	// The hint is enabled by default.
+	DisableContinuationHint bool `yaml:"disable-continuation-hint,omitempty" json:"disable-continuation-hint,omitempty"`
+
+	// TokenRefreshSeconds controls the background cookie auto-refresh interval in seconds.
+	// When unset or <= 0, defaults to 540 seconds.
+	TokenRefreshSeconds int `yaml:"token-refresh-seconds" json:"token-refresh-seconds"`
+}
+
+// RemoteManagement holds management API configuration under 'remote-management'.
+type RemoteManagement struct {
+	// AllowRemote toggles remote (non-localhost) access to management API.
+	AllowRemote bool `yaml:"allow-remote"`
+	// SecretKey is the management key (plaintext or bcrypt hashed). YAML key intentionally 'secret-key'.
+	SecretKey string `yaml:"secret-key"`
 }
 
 // QuotaExceeded defines the behavior when API quota limits are exceeded.
 // It provides configuration options for automatic failover mechanisms.
 type QuotaExceeded struct {
 	// SwitchProject indicates whether to automatically switch to another project when a quota is exceeded.
-	SwitchProject bool `yaml:"switch-project"`
+	SwitchProject bool `yaml:"switch-project" json:"switch-project"`
 
 	// SwitchPreviewModel indicates whether to automatically switch to a preview model when a quota is exceeded.
-	SwitchPreviewModel bool `yaml:"switch-preview-model"`
+	SwitchPreviewModel bool `yaml:"switch-preview-model" json:"switch-preview-model"`
 }
 
 // ClaudeKey represents the configuration for a Claude API key,
 // including the API key itself and an optional base URL for the API endpoint.
 type ClaudeKey struct {
 	// APIKey is the authentication key for accessing Claude API services.
-	APIKey string `yaml:"api-key"`
+	APIKey string `yaml:"api-key" json:"api-key"`
 
 	// BaseURL is the base URL for the Claude API endpoint.
 	// If empty, the default Claude API URL will be used.
-	BaseURL string `yaml:"base-url"`
+	BaseURL string `yaml:"base-url" json:"base-url"`
+}
+
+// CodexKey represents the configuration for a Codex API key,
+// including the API key itself and an optional base URL for the API endpoint.
+type CodexKey struct {
+	// APIKey is the authentication key for accessing Codex API services.
+	APIKey string `yaml:"api-key" json:"api-key"`
+
+	// BaseURL is the base URL for the Codex API endpoint.
+	// If empty, the default Codex API URL will be used.
+	BaseURL string `yaml:"base-url" json:"base-url"`
 }
 
 // OpenAICompatibility represents the configuration for OpenAI API compatibility
 // with external providers, allowing model aliases to be routed through OpenAI API format.
 type OpenAICompatibility struct {
 	// Name is the identifier for this OpenAI compatibility configuration.
-	Name string `yaml:"name"`
+	Name string `yaml:"name" json:"name"`
 
 	// BaseURL is the base URL for the external OpenAI-compatible API endpoint.
-	BaseURL string `yaml:"base-url"`
+	BaseURL string `yaml:"base-url" json:"base-url"`
 
 	// APIKeys are the authentication keys for accessing the external API services.
-	APIKeys []string `yaml:"api-keys"`
+	APIKeys []string `yaml:"api-keys" json:"api-keys"`
 
 	// Models defines the model configurations including aliases for routing.
-	Models []OpenAICompatibilityModel `yaml:"models"`
+	Models []OpenAICompatibilityModel `yaml:"models" json:"models"`
 }
 
 // OpenAICompatibilityModel represents a model configuration for OpenAI compatibility,
 // including the actual model name and its alias for API routing.
 type OpenAICompatibilityModel struct {
 	// Name is the actual model name used by the external provider.
-	Name string `yaml:"name"`
+	Name string `yaml:"name" json:"name"`
 
 	// Alias is the model name alias that clients will use to reference this model.
-	Alias string `yaml:"alias"`
+	Alias string `yaml:"alias" json:"alias"`
 }
 
 // LoadConfig reads a YAML configuration file from the given path,
@@ -116,10 +176,298 @@ func LoadConfig(configFile string) (*Config, error) {
 
 	// Unmarshal the YAML data into the Config struct.
 	var config Config
+	// Set defaults before unmarshal so that absent keys keep defaults.
+	config.GeminiWeb.Context = true
 	if err = yaml.Unmarshal(data, &config); err != nil {
 		return nil, fmt.Errorf("failed to parse config file: %w", err)
 	}
 
+	// Hash remote management key if plaintext is detected (nested)
+	// We consider a value to be already hashed if it looks like a bcrypt hash ($2a$, $2b$, or $2y$ prefix).
+	if config.RemoteManagement.SecretKey != "" && !looksLikeBcrypt(config.RemoteManagement.SecretKey) {
+		hashed, errHash := hashSecret(config.RemoteManagement.SecretKey)
+		if errHash != nil {
+			return nil, fmt.Errorf("failed to hash remote management key: %w", errHash)
+		}
+		config.RemoteManagement.SecretKey = hashed
+
+		// Persist the hashed value back to the config file to avoid re-hashing on next startup.
+		// Preserve YAML comments and ordering; update only the nested key.
+		_ = SaveConfigPreserveCommentsUpdateNestedScalar(configFile, []string{"remote-management", "secret-key"}, hashed)
+	}
+
 	// Return the populated configuration struct.
 	return &config, nil
 }
+
+// looksLikeBcrypt returns true if the provided string appears to be a bcrypt hash.
+func looksLikeBcrypt(s string) bool {
+	return len(s) > 4 && (s[:4] == "$2a$" || s[:4] == "$2b$" || s[:4] == "$2y$")
+}
+
+// hashSecret hashes the given secret using bcrypt.
+func hashSecret(secret string) (string, error) {
+	// Use default cost for simplicity.
+	hashedBytes, err := bcrypt.GenerateFromPassword([]byte(secret), bcrypt.DefaultCost)
+	if err != nil {
+		return "", err
+	}
+	return string(hashedBytes), nil
+}
+
+// SaveConfigPreserveComments writes the config back to YAML while preserving existing comments
+// and key ordering by loading the original file into a yaml.Node tree and updating values in-place.
+func SaveConfigPreserveComments(configFile string, cfg *Config) error {
+	// Load original YAML as a node tree to preserve comments and ordering.
+	data, err := os.ReadFile(configFile)
+	if err != nil {
+		return err
+	}
+
+	var original yaml.Node
+	if err = yaml.Unmarshal(data, &original); err != nil {
+		return err
+	}
+	if original.Kind != yaml.DocumentNode || len(original.Content) == 0 {
+		return fmt.Errorf("invalid yaml document structure")
+	}
+	if original.Content[0] == nil || original.Content[0].Kind != yaml.MappingNode {
+		return fmt.Errorf("expected root mapping node")
+	}
+
+	// Marshal the current cfg to YAML, then unmarshal to a yaml.Node we can merge from.
+	rendered, err := yaml.Marshal(cfg)
+	if err != nil {
+		return err
+	}
+	var generated yaml.Node
+	if err = yaml.Unmarshal(rendered, &generated); err != nil {
+		return err
+	}
+	if generated.Kind != yaml.DocumentNode || len(generated.Content) == 0 || generated.Content[0] == nil {
+		return fmt.Errorf("invalid generated yaml structure")
+	}
+	if generated.Content[0].Kind != yaml.MappingNode {
+		return fmt.Errorf("expected generated root mapping node")
+	}
+
+	// Merge generated into original in-place, preserving comments/order of existing nodes.
+	mergeMappingPreserve(original.Content[0], generated.Content[0])
+
+	// Write back.
+	f, err := os.Create(configFile)
+	if err != nil {
+		return err
+	}
+	defer func() { _ = f.Close() }()
+	enc := yaml.NewEncoder(f)
+	enc.SetIndent(2)
+	if err = enc.Encode(&original); err != nil {
+		_ = enc.Close()
+		return err
+	}
+	return enc.Close()
+}
+
+// SaveConfigPreserveCommentsUpdateNestedScalar updates a nested scalar key path like ["a","b"]
+// while preserving comments and positions.
+func SaveConfigPreserveCommentsUpdateNestedScalar(configFile string, path []string, value string) error {
+	data, err := os.ReadFile(configFile)
+	if err != nil {
+		return err
+	}
+	var root yaml.Node
+	if err = yaml.Unmarshal(data, &root); err != nil {
+		return err
+	}
+	if root.Kind != yaml.DocumentNode || len(root.Content) == 0 {
+		return fmt.Errorf("invalid yaml document structure")
+	}
+	node := root.Content[0]
+	// descend mapping nodes following path
+	for i, key := range path {
+		if i == len(path)-1 {
+			// set final scalar
+			v := getOrCreateMapValue(node, key)
+			v.Kind = yaml.ScalarNode
+			v.Tag = "!!str"
+			v.Value = value
+		} else {
+			next := getOrCreateMapValue(node, key)
+			if next.Kind != yaml.MappingNode {
+				next.Kind = yaml.MappingNode
+				next.Tag = "!!map"
+			}
+			node = next
+		}
+	}
+	f, err := os.Create(configFile)
+	if err != nil {
+		return err
+	}
+	defer func() { _ = f.Close() }()
+	enc := yaml.NewEncoder(f)
+	enc.SetIndent(2)
+	if err = enc.Encode(&root); err != nil {
+		_ = enc.Close()
+		return err
+	}
+	return enc.Close()
+}
+
+// getOrCreateMapValue finds the value node for a given key in a mapping node.
+// If not found, it appends a new key/value pair and returns the new value node.
+func getOrCreateMapValue(mapNode *yaml.Node, key string) *yaml.Node {
+	if mapNode.Kind != yaml.MappingNode {
+		mapNode.Kind = yaml.MappingNode
+		mapNode.Tag = "!!map"
+		mapNode.Content = nil
+	}
+	for i := 0; i+1 < len(mapNode.Content); i += 2 {
+		k := mapNode.Content[i]
+		if k.Value == key {
+			return mapNode.Content[i+1]
+		}
+	}
+	// append new key/value
+	mapNode.Content = append(mapNode.Content, &yaml.Node{Kind: yaml.ScalarNode, Tag: "!!str", Value: key})
+	val := &yaml.Node{Kind: yaml.ScalarNode, Tag: "!!str", Value: ""}
+	mapNode.Content = append(mapNode.Content, val)
+	return val
+}
+
+// mergeMappingPreserve merges keys from src into dst mapping node while preserving
+// key order and comments of existing keys in dst. Unknown keys from src are appended
+// to dst at the end, copying their node structure from src.
+func mergeMappingPreserve(dst, src *yaml.Node) {
+	if dst == nil || src == nil {
+		return
+	}
+	if dst.Kind != yaml.MappingNode || src.Kind != yaml.MappingNode {
+		// If kinds do not match, prefer replacing dst with src semantics in-place
+		// but keep dst node object to preserve any attached comments at the parent level.
+		copyNodeShallow(dst, src)
+		return
+	}
+	// Build a lookup of existing keys in dst
+	for i := 0; i+1 < len(src.Content); i += 2 {
+		sk := src.Content[i]
+		sv := src.Content[i+1]
+		idx := findMapKeyIndex(dst, sk.Value)
+		if idx >= 0 {
+			// Merge into existing value node
+			dv := dst.Content[idx+1]
+			mergeNodePreserve(dv, sv)
+		} else {
+			// Append new key/value pair by deep-copying from src
+			dst.Content = append(dst.Content, deepCopyNode(sk), deepCopyNode(sv))
+		}
+	}
+}
+
+// mergeNodePreserve merges src into dst for scalars, mappings and sequences while
+// reusing destination nodes to keep comments and anchors. For sequences, it updates
+// in-place by index.
+func mergeNodePreserve(dst, src *yaml.Node) {
+	if dst == nil || src == nil {
+		return
+	}
+	switch src.Kind {
+	case yaml.MappingNode:
+		if dst.Kind != yaml.MappingNode {
+			copyNodeShallow(dst, src)
+		}
+		mergeMappingPreserve(dst, src)
+	case yaml.SequenceNode:
+		// Preserve explicit null style if dst was null and src is empty sequence
+		if dst.Kind == yaml.ScalarNode && dst.Tag == "!!null" && len(src.Content) == 0 {
+			// Keep as null to preserve original style
+			return
+		}
+		if dst.Kind != yaml.SequenceNode {
+			dst.Kind = yaml.SequenceNode
+			dst.Tag = "!!seq"
+			dst.Content = nil
+		}
+		// Update elements in place
+		minContent := len(dst.Content)
+		if len(src.Content) < minContent {
+			minContent = len(src.Content)
+		}
+		for i := 0; i < minContent; i++ {
+			if dst.Content[i] == nil {
+				dst.Content[i] = deepCopyNode(src.Content[i])
+				continue
+			}
+			mergeNodePreserve(dst.Content[i], src.Content[i])
+		}
+		// Append any extra items from src
+		for i := len(dst.Content); i < len(src.Content); i++ {
+			dst.Content = append(dst.Content, deepCopyNode(src.Content[i]))
+		}
+		// Truncate if dst has extra items not in src
+		if len(src.Content) < len(dst.Content) {
+			dst.Content = dst.Content[:len(src.Content)]
+		}
+	case yaml.ScalarNode, yaml.AliasNode:
+		// For scalars, update Tag and Value but keep Style from dst to preserve quoting
+		dst.Kind = src.Kind
+		dst.Tag = src.Tag
+		dst.Value = src.Value
+		// Keep dst.Style as-is intentionally
+	case 0:
+		// Unknown/empty kind; do nothing
+	default:
+		// Fallback: replace shallowly
+		copyNodeShallow(dst, src)
+	}
+}
+
+// findMapKeyIndex returns the index of key node in dst mapping (index of key, not value).
+// Returns -1 when not found.
+func findMapKeyIndex(mapNode *yaml.Node, key string) int {
+	if mapNode == nil || mapNode.Kind != yaml.MappingNode {
+		return -1
+	}
+	for i := 0; i+1 < len(mapNode.Content); i += 2 {
+		if mapNode.Content[i] != nil && mapNode.Content[i].Value == key {
+			return i
+		}
+	}
+	return -1
+}
+
+// deepCopyNode creates a deep copy of a yaml.Node graph.
+func deepCopyNode(n *yaml.Node) *yaml.Node {
+	if n == nil {
+		return nil
+	}
+	cp := *n
+	if len(n.Content) > 0 {
+		cp.Content = make([]*yaml.Node, len(n.Content))
+		for i := range n.Content {
+			cp.Content[i] = deepCopyNode(n.Content[i])
+		}
+	}
+	return &cp
+}
+
+// copyNodeShallow copies type/tag/value and resets content to match src, but
+// keeps the same destination node pointer to preserve parent relations/comments.
+func copyNodeShallow(dst, src *yaml.Node) {
+	if dst == nil || src == nil {
+		return
+	}
+	dst.Kind = src.Kind
+	dst.Tag = src.Tag
+	dst.Value = src.Value
+	// Replace content with deep copy from src
+	if len(src.Content) > 0 {
+		dst.Content = make([]*yaml.Node, len(src.Content))
+		for i := range src.Content {
+			dst.Content[i] = deepCopyNode(src.Content[i])
+		}
+	} else {
+		dst.Content = nil
+	}
+}

internal/constant/constant.go

@@ -1,10 +1,10 @@
 package constant
 
 const (
-	GEMINI               = "gemini"
-	GEMINICLI            = "gemini-cli"
-	CODEX                = "codex"
-	CLAUDE               = "claude"
-	OPENAI               = "openai"
-	OPENAI_COMPATIBILITY = "openai-compatibility"
+	GEMINI          = "gemini"
+	GEMINICLI       = "gemini-cli"
+	CODEX           = "codex"
+	CLAUDE          = "claude"
+	OPENAI          = "openai"
+	OPENAI_RESPONSE = "openai-response"
 )

internal/interfaces/client.go

@@ -51,4 +51,27 @@ type Client interface {
 
 	// Provider returns the name of the AI service provider (e.g., "gemini", "claude").
 	Provider() string
+
+	// RefreshTokens refreshes the access tokens if needed
+	RefreshTokens(ctx context.Context) error
+
+	// IsAvailable returns true if the client is available for use.
+	IsAvailable() bool
+
+	// SetUnavailable sets the client to unavailable.
+	SetUnavailable()
 }
+
+// UnregisterReason describes the context for unregistering a client instance.
+type UnregisterReason string
+
+const (
+	// UnregisterReasonReload indicates a full reload is replacing the client.
+	UnregisterReasonReload UnregisterReason = "reload"
+	// UnregisterReasonShutdown indicates the service is shutting down.
+	UnregisterReasonShutdown UnregisterReason = "shutdown"
+	// UnregisterReasonAuthFileRemoved indicates the underlying auth file was deleted.
+	UnregisterReasonAuthFileRemoved UnregisterReason = "auth-file-removed"
+	// UnregisterReasonAuthFileUpdated indicates the auth file content was modified.
+	UnregisterReasonAuthFileUpdated UnregisterReason = "auth-file-updated"
+)

internal/interfaces/types.go

@@ -28,7 +28,7 @@ type TranslateRequestFunc func(string, []byte, bool) []byte
 //
 // Returns:
 //   - []string: An array of translated response strings
-type TranslateResponseFunc func(ctx context.Context, modelName string, rawJSON []byte, param *any) []string
+type TranslateResponseFunc func(ctx context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, param *any) []string
 
 // TranslateResponseNonStreamFunc defines a function type for translating non-streaming API responses.
 // It processes response data and returns a single translated response string.
@@ -41,7 +41,7 @@ type TranslateResponseFunc func(ctx context.Context, modelName string, rawJSON [
 //
 // Returns:
 //   - string: A single translated response string
-type TranslateResponseNonStreamFunc func(ctx context.Context, modelName string, rawJSON []byte, param *any) string
+type TranslateResponseNonStreamFunc func(ctx context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, param *any) string
 
 // TranslateResponse contains both streaming and non-streaming response translation functions.
 // This structure allows clients to handle both types of API responses appropriately.

internal/logging/request_logger.go

@@ -14,6 +14,8 @@ import (
 	"regexp"
 	"strings"
 	"time"
+
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
 )
 
 // RequestLogger defines the interface for logging HTTP requests and responses.
@@ -34,7 +36,7 @@ type RequestLogger interface {
 	//
 	// Returns:
 	//   - error: An error if logging fails, nil otherwise
-	LogRequest(url, method string, requestHeaders map[string][]string, body []byte, statusCode int, responseHeaders map[string][]string, response, apiRequest, apiResponse []byte) error
+	LogRequest(url, method string, requestHeaders map[string][]string, body []byte, statusCode int, responseHeaders map[string][]string, response, apiRequest, apiResponse []byte, apiResponseErrors []*interfaces.ErrorMessage) error
 
 	// LogStreamingRequest initiates logging for a streaming request and returns a writer for chunks.
 	//
@@ -96,11 +98,20 @@ type FileRequestLogger struct {
 //
 // Parameters:
 //   - enabled: Whether request logging should be enabled
-//   - logsDir: The directory where log files should be stored
+//   - logsDir: The directory where log files should be stored (can be relative)
+//   - configDir: The directory of the configuration file; when logsDir is
+//     relative, it will be resolved relative to this directory
 //
 // Returns:
 //   - *FileRequestLogger: A new file-based request logger instance
-func NewFileRequestLogger(enabled bool, logsDir string) *FileRequestLogger {
+func NewFileRequestLogger(enabled bool, logsDir string, configDir string) *FileRequestLogger {
+	// Resolve logsDir relative to the configuration file directory when it's not absolute.
+	if !filepath.IsAbs(logsDir) {
+		// If configDir is provided, resolve logsDir relative to it.
+		if configDir != "" {
+			logsDir = filepath.Join(configDir, logsDir)
+		}
+	}
 	return &FileRequestLogger{
 		enabled: enabled,
 		logsDir: logsDir,
@@ -139,7 +150,7 @@ func (l *FileRequestLogger) SetEnabled(enabled bool) {
 //
 // Returns:
 //   - error: An error if logging fails, nil otherwise
-func (l *FileRequestLogger) LogRequest(url, method string, requestHeaders map[string][]string, body []byte, statusCode int, responseHeaders map[string][]string, response, apiRequest, apiResponse []byte) error {
+func (l *FileRequestLogger) LogRequest(url, method string, requestHeaders map[string][]string, body []byte, statusCode int, responseHeaders map[string][]string, response, apiRequest, apiResponse []byte, apiResponseErrors []*interfaces.ErrorMessage) error {
 	if !l.enabled {
 		return nil
 	}
@@ -161,7 +172,7 @@ func (l *FileRequestLogger) LogRequest(url, method string, requestHeaders map[st
 	}
 
 	// Create log content
-	content := l.formatLogContent(url, method, requestHeaders, body, apiRequest, apiResponse, decompressedResponse, statusCode, responseHeaders)
+	content := l.formatLogContent(url, method, requestHeaders, body, apiRequest, apiResponse, decompressedResponse, statusCode, responseHeaders, apiResponseErrors)
 
 	// Write to file
 	if err = os.WriteFile(filePath, []byte(content), 0644); err != nil {
@@ -310,7 +321,7 @@ func (l *FileRequestLogger) sanitizeForFilename(path string) string {
 //
 // Returns:
 //   - string: The formatted log content
-func (l *FileRequestLogger) formatLogContent(url, method string, headers map[string][]string, body, apiRequest, apiResponse, response []byte, status int, responseHeaders map[string][]string) string {
+func (l *FileRequestLogger) formatLogContent(url, method string, headers map[string][]string, body, apiRequest, apiResponse, response []byte, status int, responseHeaders map[string][]string, apiResponseErrors []*interfaces.ErrorMessage) string {
 	var content strings.Builder
 
 	// Request info
@@ -320,6 +331,13 @@ func (l *FileRequestLogger) formatLogContent(url, method string, headers map[str
 	content.Write(apiRequest)
 	content.WriteString("\n\n")
 
+	for i := 0; i < len(apiResponseErrors); i++ {
+		content.WriteString("=== API ERROR RESPONSE ===\n")
+		content.WriteString(fmt.Sprintf("HTTP Status: %d\n", apiResponseErrors[i].StatusCode))
+		content.WriteString(apiResponseErrors[i].Error.Error())
+		content.WriteString("\n\n")
+	}
+
 	content.WriteString("=== API RESPONSE ===\n")
 	content.Write(apiResponse)
 	content.WriteString("\n\n")

internal/misc/codex_instructions.go

@@ -9,5 +9,15 @@ import _ "embed"
 // which is embedded into the application binary at compile time. This variable
 // contains instructional text used for Codex-related operations and model guidance.
 //
-//go:embed codex_instructions.txt
-var CodexInstructions string
+//go:embed gpt_5_instructions.txt
+var GPT5Instructions string
+
+//go:embed gpt_5_codex_instructions.txt
+var GPT5CodexInstructions string
+
+func CodexInstructions(modelName string) string {
+	if modelName == "gpt-5-codex" {
+		return GPT5CodexInstructions
+	}
+	return GPT5Instructions
+}

internal/misc/credentials.go

@@ -0,0 +1,24 @@
+package misc
+
+import (
+	"path/filepath"
+	"strings"
+
+	log "github.com/sirupsen/logrus"
+)
+
+var credentialSeparator = strings.Repeat("-", 70)
+
+// LogSavingCredentials emits a consistent log message when persisting auth material.
+func LogSavingCredentials(path string) {
+	if path == "" {
+		return
+	}
+	// Use filepath.Clean so logs remain stable even if callers pass redundant separators.
+	log.Infof("Saving credentials to %s", filepath.Clean(path))
+}
+
+// LogCredentialSeparator adds a visual separator to group auth/key processing logs.
+func LogCredentialSeparator() {
+	log.Info(credentialSeparator)
+}

internal/misc/oauth.go

@@ -0,0 +1,21 @@
+package misc
+
+import (
+	"crypto/rand"
+	"encoding/hex"
+	"fmt"
+)
+
+// GenerateRandomState generates a cryptographically secure random state parameter
+// for OAuth2 flows to prevent CSRF attacks.
+//
+// Returns:
+//   - string: A hexadecimal encoded random state string
+//   - error: An error if the random generation fails, nil otherwise
+func GenerateRandomState() (string, error) {
+	bytes := make([]byte, 16)
+	if _, err := rand.Read(bytes); err != nil {
+		return "", fmt.Errorf("failed to generate random bytes: %w", err)
+	}
+	return hex.EncodeToString(bytes), nil
+}

internal/registry/model_definitions.go

@@ -130,6 +130,20 @@ func GetGeminiCLIModels() []*ModelInfo {
 			OutputTokenLimit:           65536,
 			SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
 		},
+		{
+			ID:                         "gemini-2.5-flash-lite",
+			Object:                     "model",
+			Created:                    time.Now().Unix(),
+			OwnedBy:                    "google",
+			Type:                       "gemini",
+			Name:                       "models/gemini-2.5-flash-lite",
+			Version:                    "2.5",
+			DisplayName:                "Gemini 2.5 Flash Lite",
+			Description:                "Our smallest and most cost effective model, built for at scale usage.",
+			InputTokenLimit:            1048576,
+			OutputTokenLimit:           65536,
+			SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
+		},
 	}
 }
 
@@ -149,6 +163,110 @@ func GetOpenAIModels() []*ModelInfo {
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
 		},
+		{
+			ID:                  "gpt-5-minimal",
+			Object:              "model",
+			Created:             time.Now().Unix(),
+			OwnedBy:             "openai",
+			Type:                "openai",
+			Version:             "gpt-5-2025-08-07",
+			DisplayName:         "GPT 5 Minimal",
+			Description:         "Stable version of GPT 5, The best model for coding and agentic tasks across domains.",
+			ContextLength:       400000,
+			MaxCompletionTokens: 128000,
+			SupportedParameters: []string{"tools"},
+		},
+		{
+			ID:                  "gpt-5-low",
+			Object:              "model",
+			Created:             time.Now().Unix(),
+			OwnedBy:             "openai",
+			Type:                "openai",
+			Version:             "gpt-5-2025-08-07",
+			DisplayName:         "GPT 5 Low",
+			Description:         "Stable version of GPT 5, The best model for coding and agentic tasks across domains.",
+			ContextLength:       400000,
+			MaxCompletionTokens: 128000,
+			SupportedParameters: []string{"tools"},
+		},
+		{
+			ID:                  "gpt-5-medium",
+			Object:              "model",
+			Created:             time.Now().Unix(),
+			OwnedBy:             "openai",
+			Type:                "openai",
+			Version:             "gpt-5-2025-08-07",
+			DisplayName:         "GPT 5 Medium",
+			Description:         "Stable version of GPT 5, The best model for coding and agentic tasks across domains.",
+			ContextLength:       400000,
+			MaxCompletionTokens: 128000,
+			SupportedParameters: []string{"tools"},
+		},
+		{
+			ID:                  "gpt-5-high",
+			Object:              "model",
+			Created:             time.Now().Unix(),
+			OwnedBy:             "openai",
+			Type:                "openai",
+			Version:             "gpt-5-2025-08-07",
+			DisplayName:         "GPT 5 High",
+			Description:         "Stable version of GPT 5, The best model for coding and agentic tasks across domains.",
+			ContextLength:       400000,
+			MaxCompletionTokens: 128000,
+			SupportedParameters: []string{"tools"},
+		},
+		{
+			ID:                  "gpt-5-codex",
+			Object:              "model",
+			Created:             time.Now().Unix(),
+			OwnedBy:             "openai",
+			Type:                "openai",
+			Version:             "gpt-5-2025-09-15",
+			DisplayName:         "GPT 5 Codex",
+			Description:         "Stable version of GPT 5 Codex, The best model for coding and agentic tasks across domains.",
+			ContextLength:       400000,
+			MaxCompletionTokens: 128000,
+			SupportedParameters: []string{"tools"},
+		},
+		{
+			ID:                  "gpt-5-codex-low",
+			Object:              "model",
+			Created:             time.Now().Unix(),
+			OwnedBy:             "openai",
+			Type:                "openai",
+			Version:             "gpt-5-2025-09-15",
+			DisplayName:         "GPT 5 Codex Low",
+			Description:         "Stable version of GPT 5 Codex, The best model for coding and agentic tasks across domains.",
+			ContextLength:       400000,
+			MaxCompletionTokens: 128000,
+			SupportedParameters: []string{"tools"},
+		},
+		{
+			ID:                  "gpt-5-codex-medium",
+			Object:              "model",
+			Created:             time.Now().Unix(),
+			OwnedBy:             "openai",
+			Type:                "openai",
+			Version:             "gpt-5-2025-09-15",
+			DisplayName:         "GPT 5 Codex Medium",
+			Description:         "Stable version of GPT 5 Codex, The best model for coding and agentic tasks across domains.",
+			ContextLength:       400000,
+			MaxCompletionTokens: 128000,
+			SupportedParameters: []string{"tools"},
+		},
+		{
+			ID:                  "gpt-5-codex-high",
+			Object:              "model",
+			Created:             time.Now().Unix(),
+			OwnedBy:             "openai",
+			Type:                "openai",
+			Version:             "gpt-5-2025-09-15",
+			DisplayName:         "GPT 5 Codex High",
+			Description:         "Stable version of GPT 5 Codex, The best model for coding and agentic tasks across domains.",
+			ContextLength:       400000,
+			MaxCompletionTokens: 128000,
+			SupportedParameters: []string{"tools"},
+		},
 		{
 			ID:                  "codex-mini-latest",
 			Object:              "model",

internal/registry/model_registry.go

@@ -185,7 +185,7 @@ func (r *ModelRegistry) ClearModelQuotaExceeded(clientID, modelID string) {
 
 	if registration, exists := r.models[modelID]; exists {
 		delete(registration.QuotaExceededClients, clientID)
-		log.Debugf("Cleared quota exceeded status for model %s and client %s", modelID, clientID)
+		// log.Debugf("Cleared quota exceeded status for model %s and client %s", modelID, clientID)
 	}
 }
 

internal/translator/claude/gemini-cli/claude_gemini-cli_request.go

@@ -6,7 +6,9 @@
 package geminiCLI
 
 import (
-	. "github.com/luispater/CLIProxyAPI/internal/translator/claude/gemini"
+	"bytes"
+
+	. "github.com/luispater/CLIProxyAPI/v5/internal/translator/claude/gemini"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
 )
@@ -27,7 +29,9 @@ import (
 //
 // Returns:
 //   - []byte: The transformed request data in Claude Code API format
-func ConvertGeminiCLIRequestToClaude(modelName string, rawJSON []byte, stream bool) []byte {
+func ConvertGeminiCLIRequestToClaude(modelName string, inputRawJSON []byte, stream bool) []byte {
+	rawJSON := bytes.Clone(inputRawJSON)
+
 	modelResult := gjson.GetBytes(rawJSON, "model")
 	// Extract the inner request object and promote it to the top level
 	rawJSON = []byte(gjson.GetBytes(rawJSON, "request").Raw)

internal/translator/claude/gemini-cli/claude_gemini-cli_response.go

@@ -7,7 +7,7 @@ package geminiCLI
 import (
 	"context"
 
-	. "github.com/luispater/CLIProxyAPI/internal/translator/claude/gemini"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/translator/claude/gemini"
 	"github.com/tidwall/sjson"
 )
 
@@ -24,8 +24,8 @@ import (
 //
 // Returns:
 //   - []string: A slice of strings, each containing a Gemini-compatible JSON response wrapped in a response object
-func ConvertClaudeResponseToGeminiCLI(ctx context.Context, modelName string, rawJSON []byte, param *any) []string {
-	outputs := ConvertClaudeResponseToGemini(ctx, modelName, rawJSON, param)
+func ConvertClaudeResponseToGeminiCLI(ctx context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, param *any) []string {
+	outputs := ConvertClaudeResponseToGemini(ctx, modelName, originalRequestRawJSON, requestRawJSON, rawJSON, param)
 	// Wrap each converted response in a "response" object to match Gemini CLI API structure
 	newOutputs := make([]string, 0)
 	for i := 0; i < len(outputs); i++ {
@@ -48,8 +48,8 @@ func ConvertClaudeResponseToGeminiCLI(ctx context.Context, modelName string, raw
 //
 // Returns:
 //   - string: A Gemini-compatible JSON response wrapped in a response object
-func ConvertClaudeResponseToGeminiCLINonStream(ctx context.Context, modelName string, rawJSON []byte, param *any) string {
-	strJSON := ConvertClaudeResponseToGeminiNonStream(ctx, modelName, rawJSON, param)
+func ConvertClaudeResponseToGeminiCLINonStream(ctx context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, param *any) string {
+	strJSON := ConvertClaudeResponseToGeminiNonStream(ctx, modelName, originalRequestRawJSON, requestRawJSON, rawJSON, param)
 	// Wrap the converted response in a "response" object to match Gemini CLI API structure
 	json := `{"response": {}}`
 	strJSON, _ = sjson.SetRaw(json, "response", strJSON)

internal/translator/claude/gemini-cli/init.go

@@ -1,9 +1,9 @@
 package geminiCLI
 
 import (
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/translator/translator"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/translator/translator"
 )
 
 func init() {

internal/translator/claude/gemini/claude_gemini_request.go

@@ -6,12 +6,13 @@
 package gemini
 
 import (
+	"bytes"
 	"crypto/rand"
 	"fmt"
 	"math/big"
 	"strings"
 
-	"github.com/luispater/CLIProxyAPI/internal/util"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
 )
@@ -34,7 +35,8 @@ import (
 //
 // Returns:
 //   - []byte: The transformed request data in Claude Code API format
-func ConvertGeminiRequestToClaude(modelName string, rawJSON []byte, stream bool) []byte {
+func ConvertGeminiRequestToClaude(modelName string, inputRawJSON []byte, stream bool) []byte {
+	rawJSON := bytes.Clone(inputRawJSON)
 	// Base Claude Code API template with default max_tokens value
 	out := `{"model":"","max_tokens":32000,"messages":[]}`
 
@@ -87,6 +89,17 @@ func ConvertGeminiRequestToClaude(modelName string, rawJSON []byte, stream bool)
 				out, _ = sjson.Set(out, "stop_sequences", stopSequences)
 			}
 		}
+		// Include thoughts configuration for reasoning process visibility
+		if thinkingConfig := genConfig.Get("thinkingConfig"); thinkingConfig.Exists() && thinkingConfig.IsObject() {
+			if includeThoughts := thinkingConfig.Get("include_thoughts"); includeThoughts.Exists() {
+				if includeThoughts.Type == gjson.True {
+					out, _ = sjson.Set(out, "thinking.type", "enabled")
+					if thinkingBudget := thinkingConfig.Get("thinkingBudget"); thinkingBudget.Exists() {
+						out, _ = sjson.Set(out, "thinking.budget_tokens", thinkingBudget.Int())
+					}
+				}
+			}
+		}
 	}
 
 	// System instruction conversion to Claude Code format

internal/translator/claude/gemini/claude_gemini_response.go

@@ -52,7 +52,7 @@ type ConvertAnthropicResponseToGeminiParams struct {
 //
 // Returns:
 //   - []string: A slice of strings, each containing a Gemini-compatible JSON response
-func ConvertClaudeResponseToGemini(_ context.Context, modelName string, rawJSON []byte, param *any) []string {
+func ConvertClaudeResponseToGemini(_ context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, param *any) []string {
 	if *param == nil {
 		*param = &ConvertAnthropicResponseToGeminiParams{
 			Model:      modelName,
@@ -128,7 +128,7 @@ func ConvertClaudeResponseToGemini(_ context.Context, modelName string, rawJSON
 				}
 			case "thinking_delta":
 				// Thinking/reasoning content delta for models with reasoning capabilities
-				if text := delta.Get("text"); text.Exists() && text.String() != "" {
+				if text := delta.Get("thinking"); text.Exists() && text.String() != "" {
 					thinkingPart := `{"thought":true,"text":""}`
 					thinkingPart, _ = sjson.Set(thinkingPart, "text", text.String())
 					template, _ = sjson.SetRaw(template, "candidates.0.content.parts.-1", thinkingPart)
@@ -320,7 +320,7 @@ func convertMapToJSON(m map[string]interface{}) string {
 //
 // Returns:
 //   - string: A Gemini-compatible JSON response containing all message content and metadata
-func ConvertClaudeResponseToGeminiNonStream(_ context.Context, modelName string, rawJSON []byte, _ *any) string {
+func ConvertClaudeResponseToGeminiNonStream(_ context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, _ *any) string {
 	// Base Gemini response template for non-streaming with default values
 	template := `{"candidates":[{"content":{"role":"model","parts":[]},"finishReason":"STOP"}],"usageMetadata":{"trafficType":"PROVISIONED_THROUGHPUT"},"modelVersion":"","createTime":"","responseId":""}`
 
@@ -411,7 +411,7 @@ func ConvertClaudeResponseToGeminiNonStream(_ context.Context, modelName string,
 					}
 				case "thinking_delta":
 					// Process reasoning/thinking content
-					if text := delta.Get("text"); text.Exists() && text.String() != "" {
+					if text := delta.Get("thinking"); text.Exists() && text.String() != "" {
 						partJSON := `{"thought":true,"text":""}`
 						partJSON, _ = sjson.Set(partJSON, "text", text.String())
 						part := gjson.Parse(partJSON).Value().(map[string]interface{})

internal/translator/claude/gemini/init.go

@@ -1,9 +1,9 @@
 package gemini
 
 import (
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/translator/translator"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/translator/translator"
 )
 
 func init() {

internal/translator/claude/openai/chat-completions/claude_openai_request.go

@@ -3,9 +3,10 @@
 // extracting model information, system instructions, message contents, and tool declarations.
 // The package performs JSON data transformation to ensure compatibility
 // between OpenAI API format and Claude Code API's expected format.
-package openai
+package chat_completions
 
 import (
+	"bytes"
 	"crypto/rand"
 	"encoding/json"
 	"math/big"
@@ -32,12 +33,29 @@ import (
 //
 // Returns:
 //   - []byte: The transformed request data in Claude Code API format
-func ConvertOpenAIRequestToClaude(modelName string, rawJSON []byte, stream bool) []byte {
+func ConvertOpenAIRequestToClaude(modelName string, inputRawJSON []byte, stream bool) []byte {
+	rawJSON := bytes.Clone(inputRawJSON)
+
 	// Base Claude Code API template with default max_tokens value
 	out := `{"model":"","max_tokens":32000,"messages":[]}`
 
 	root := gjson.ParseBytes(rawJSON)
 
+	if v := root.Get("reasoning_effort"); v.Exists() {
+		out, _ = sjson.Set(out, "thinking.type", "enabled")
+
+		switch v.String() {
+		case "none":
+			out, _ = sjson.Set(out, "thinking.type", "disabled")
+		case "low":
+			out, _ = sjson.Set(out, "thinking.budget_tokens", 1024)
+		case "medium":
+			out, _ = sjson.Set(out, "thinking.budget_tokens", 8192)
+		case "high":
+			out, _ = sjson.Set(out, "thinking.budget_tokens", 24576)
+		}
+	}
+
 	// Helper for generating tool call IDs in the form: toolu_<alphanum>
 	// This ensures unique identifiers for tool calls in the Claude Code format
 	genToolCallID := func() string {
@@ -244,7 +262,7 @@ func ConvertOpenAIRequestToClaude(modelName string, rawJSON []byte, stream bool)
 	}
 
 	// Tools mapping: OpenAI tools -> Claude Code tools
-	if tools := root.Get("tools"); tools.Exists() && tools.IsArray() {
+	if tools := root.Get("tools"); tools.Exists() && tools.IsArray() && len(tools.Array()) > 0 {
 		var anthropicTools []interface{}
 		tools.ForEach(func(_, tool gjson.Result) bool {
 			if tool.Get("type").String() == "function" {

internal/translator/claude/openai/chat-completions/claude_openai_response.go

@@ -3,7 +3,7 @@
 // JSON format, transforming streaming events and non-streaming responses into the format
 // expected by OpenAI API clients. It supports both streaming and non-streaming modes,
 // handling text content, tool calls, reasoning content, and usage metadata appropriately.
-package openai
+package chat_completions
 
 import (
 	"bufio"
@@ -50,7 +50,7 @@ type ToolCallAccumulator struct {
 //
 // Returns:
 //   - []string: A slice of strings, each containing an OpenAI-compatible JSON response
-func ConvertClaudeResponseToOpenAI(_ context.Context, modelName string, rawJSON []byte, param *any) []string {
+func ConvertClaudeResponseToOpenAI(_ context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, param *any) []string {
 	if *param == nil {
 		*param = &ConvertAnthropicResponseToOpenAIParams{
 			CreatedAt:    0,
@@ -128,10 +128,11 @@ func ConvertClaudeResponseToOpenAI(_ context.Context, modelName string, rawJSON
 				return []string{}
 			}
 		}
-		return []string{template}
+		return []string{}
 
 	case "content_block_delta":
 		// Handle content delta (text, tool use arguments, or reasoning content)
+		hasContent := false
 		if delta := root.Get("delta"); delta.Exists() {
 			deltaType := delta.Get("type").String()
 
@@ -140,8 +141,14 @@ func ConvertClaudeResponseToOpenAI(_ context.Context, modelName string, rawJSON
 				// Text content delta - send incremental text updates
 				if text := delta.Get("text"); text.Exists() {
 					template, _ = sjson.Set(template, "choices.0.delta.content", text.String())
+					hasContent = true
+				}
+			case "thinking_delta":
+				// Accumulate reasoning/thinking content
+				if thinking := delta.Get("thinking"); thinking.Exists() {
+					template, _ = sjson.Set(template, "choices.0.delta.reasoning_content", thinking.String())
+					hasContent = true
 				}
-
 			case "input_json_delta":
 				// Tool use input delta - accumulate arguments for tool calls
 				if partialJSON := delta.Get("partial_json"); partialJSON.Exists() {
@@ -156,7 +163,11 @@ func ConvertClaudeResponseToOpenAI(_ context.Context, modelName string, rawJSON
 				return []string{}
 			}
 		}
-		return []string{template}
+		if hasContent {
+			return []string{template}
+		} else {
+			return []string{}
+		}
 
 	case "content_block_stop":
 		// End of content block - output complete tool call if it's a tool_use block
@@ -266,7 +277,7 @@ func mapAnthropicStopReasonToOpenAI(anthropicReason string) string {
 //
 // Returns:
 //   - string: An OpenAI-compatible JSON response containing all message content and metadata
-func ConvertClaudeResponseToOpenAINonStream(_ context.Context, _ string, rawJSON []byte, _ *any) string {
+func ConvertClaudeResponseToOpenAINonStream(_ context.Context, _ string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, _ *any) string {
 	chunks := make([][]byte, 0)
 
 	scanner := bufio.NewScanner(bytes.NewReader(rawJSON))

internal/translator/claude/openai/chat-completions/init.go

@@ -1,9 +1,9 @@
-package openai
+package chat_completions
 
 import (
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/translator/translator"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/translator/translator"
 )
 
 func init() {

internal/translator/claude/openai/responses/claude_openai-responses_request.go

@@ -0,0 +1,210 @@
+package responses
+
+import (
+	"bytes"
+	"crypto/rand"
+	"math/big"
+	"strings"
+
+	"github.com/tidwall/gjson"
+	"github.com/tidwall/sjson"
+)
+
+// ConvertOpenAIResponsesRequestToClaude transforms an OpenAI Responses API request
+// into a Claude Messages API request using only gjson/sjson for JSON handling.
+// It supports:
+// - instructions -> system message
+// - input[].type==message with input_text/output_text -> user/assistant messages
+// - function_call -> assistant tool_use
+// - function_call_output -> user tool_result
+// - tools[].parameters -> tools[].input_schema
+// - max_output_tokens -> max_tokens
+// - stream passthrough via parameter
+func ConvertOpenAIResponsesRequestToClaude(modelName string, inputRawJSON []byte, stream bool) []byte {
+	rawJSON := bytes.Clone(inputRawJSON)
+
+	// Base Claude message payload
+	out := `{"model":"","max_tokens":32000,"messages":[]}`
+
+	root := gjson.ParseBytes(rawJSON)
+
+	if v := root.Get("reasoning.effort"); v.Exists() {
+		out, _ = sjson.Set(out, "thinking.type", "enabled")
+
+		switch v.String() {
+		case "none":
+			out, _ = sjson.Set(out, "thinking.type", "disabled")
+		case "minimal":
+			out, _ = sjson.Set(out, "thinking.budget_tokens", 1024)
+		case "low":
+			out, _ = sjson.Set(out, "thinking.budget_tokens", 4096)
+		case "medium":
+			out, _ = sjson.Set(out, "thinking.budget_tokens", 8192)
+		case "high":
+			out, _ = sjson.Set(out, "thinking.budget_tokens", 24576)
+		}
+	}
+
+	// Helper for generating tool call IDs when missing
+	genToolCallID := func() string {
+		const letters = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789"
+		var b strings.Builder
+		for i := 0; i < 24; i++ {
+			n, _ := rand.Int(rand.Reader, big.NewInt(int64(len(letters))))
+			b.WriteByte(letters[n.Int64()])
+		}
+		return "toolu_" + b.String()
+	}
+
+	// Model
+	out, _ = sjson.Set(out, "model", modelName)
+
+	// Max tokens
+	if mot := root.Get("max_output_tokens"); mot.Exists() {
+		out, _ = sjson.Set(out, "max_tokens", mot.Int())
+	}
+
+	// Stream
+	out, _ = sjson.Set(out, "stream", stream)
+
+	// instructions -> as a leading message (use role user for Claude API compatibility)
+	if instr := root.Get("instructions"); instr.Exists() && instr.Type == gjson.String && instr.String() != "" {
+		sysMsg := `{"role":"user","content":""}`
+		sysMsg, _ = sjson.Set(sysMsg, "content", instr.String())
+		out, _ = sjson.SetRaw(out, "messages.-1", sysMsg)
+	}
+
+	// input array processing
+	if input := root.Get("input"); input.Exists() && input.IsArray() {
+		input.ForEach(func(_, item gjson.Result) bool {
+			typ := item.Get("type").String()
+			switch typ {
+			case "message":
+				// Determine role from content type (input_text=user, output_text=assistant)
+				var role string
+				var text strings.Builder
+				if parts := item.Get("content"); parts.Exists() && parts.IsArray() {
+					parts.ForEach(func(_, part gjson.Result) bool {
+						ptype := part.Get("type").String()
+						if ptype == "input_text" || ptype == "output_text" {
+							if t := part.Get("text"); t.Exists() {
+								text.WriteString(t.String())
+							}
+							if ptype == "input_text" {
+								role = "user"
+							} else if ptype == "output_text" {
+								role = "assistant"
+							}
+						}
+						return true
+					})
+				}
+
+				// Fallback to given role if content types not decisive
+				if role == "" {
+					r := item.Get("role").String()
+					switch r {
+					case "user", "assistant", "system":
+						role = r
+					default:
+						role = "user"
+					}
+				}
+
+				if text.Len() > 0 || role == "system" {
+					msg := `{"role":"","content":""}`
+					msg, _ = sjson.Set(msg, "role", role)
+					if text.Len() > 0 {
+						msg, _ = sjson.Set(msg, "content", text.String())
+					} else {
+						msg, _ = sjson.Set(msg, "content", "")
+					}
+					out, _ = sjson.SetRaw(out, "messages.-1", msg)
+				}
+
+			case "function_call":
+				// Map to assistant tool_use
+				callID := item.Get("call_id").String()
+				if callID == "" {
+					callID = genToolCallID()
+				}
+				name := item.Get("name").String()
+				argsStr := item.Get("arguments").String()
+
+				toolUse := `{"type":"tool_use","id":"","name":"","input":{}}`
+				toolUse, _ = sjson.Set(toolUse, "id", callID)
+				toolUse, _ = sjson.Set(toolUse, "name", name)
+				if argsStr != "" && gjson.Valid(argsStr) {
+					toolUse, _ = sjson.SetRaw(toolUse, "input", argsStr)
+				}
+
+				asst := `{"role":"assistant","content":[]}`
+				asst, _ = sjson.SetRaw(asst, "content.-1", toolUse)
+				out, _ = sjson.SetRaw(out, "messages.-1", asst)
+
+			case "function_call_output":
+				// Map to user tool_result
+				callID := item.Get("call_id").String()
+				outputStr := item.Get("output").String()
+				toolResult := `{"type":"tool_result","tool_use_id":"","content":""}`
+				toolResult, _ = sjson.Set(toolResult, "tool_use_id", callID)
+				toolResult, _ = sjson.Set(toolResult, "content", outputStr)
+
+				usr := `{"role":"user","content":[]}`
+				usr, _ = sjson.SetRaw(usr, "content.-1", toolResult)
+				out, _ = sjson.SetRaw(out, "messages.-1", usr)
+			}
+			return true
+		})
+	}
+
+	// tools mapping: parameters -> input_schema
+	if tools := root.Get("tools"); tools.Exists() && tools.IsArray() {
+		toolsJSON := "[]"
+		tools.ForEach(func(_, tool gjson.Result) bool {
+			tJSON := `{"name":"","description":"","input_schema":{}}`
+			if n := tool.Get("name"); n.Exists() {
+				tJSON, _ = sjson.Set(tJSON, "name", n.String())
+			}
+			if d := tool.Get("description"); d.Exists() {
+				tJSON, _ = sjson.Set(tJSON, "description", d.String())
+			}
+
+			if params := tool.Get("parameters"); params.Exists() {
+				tJSON, _ = sjson.SetRaw(tJSON, "input_schema", params.Raw)
+			} else if params = tool.Get("parametersJsonSchema"); params.Exists() {
+				tJSON, _ = sjson.SetRaw(tJSON, "input_schema", params.Raw)
+			}
+
+			toolsJSON, _ = sjson.SetRaw(toolsJSON, "-1", tJSON)
+			return true
+		})
+		if gjson.Parse(toolsJSON).IsArray() && len(gjson.Parse(toolsJSON).Array()) > 0 {
+			out, _ = sjson.SetRaw(out, "tools", toolsJSON)
+		}
+	}
+
+	// Map tool_choice similar to Chat Completions translator (optional in docs, safe to handle)
+	if toolChoice := root.Get("tool_choice"); toolChoice.Exists() {
+		switch toolChoice.Type {
+		case gjson.String:
+			switch toolChoice.String() {
+			case "auto":
+				out, _ = sjson.Set(out, "tool_choice", map[string]interface{}{"type": "auto"})
+			case "none":
+				// Leave unset; implies no tools
+			case "required":
+				out, _ = sjson.Set(out, "tool_choice", map[string]interface{}{"type": "any"})
+			}
+		case gjson.JSON:
+			if toolChoice.Get("type").String() == "function" {
+				fn := toolChoice.Get("function.name").String()
+				out, _ = sjson.Set(out, "tool_choice", map[string]interface{}{"type": "tool", "name": fn})
+			}
+		default:
+
+		}
+	}
+
+	return []byte(out)
+}

internal/translator/claude/openai/responses/claude_openai-responses_response.go

@@ -0,0 +1,654 @@
+package responses
+
+import (
+	"bufio"
+	"bytes"
+	"context"
+	"fmt"
+	"strings"
+	"time"
+
+	"github.com/tidwall/gjson"
+	"github.com/tidwall/sjson"
+)
+
+type claudeToResponsesState struct {
+	Seq          int
+	ResponseID   string
+	CreatedAt    int64
+	CurrentMsgID string
+	CurrentFCID  string
+	InTextBlock  bool
+	InFuncBlock  bool
+	FuncArgsBuf  map[int]*strings.Builder // index -> args
+	// function call bookkeeping for output aggregation
+	FuncNames   map[int]string // index -> function name
+	FuncCallIDs map[int]string // index -> call id
+	// message text aggregation
+	TextBuf strings.Builder
+	// reasoning state
+	ReasoningActive    bool
+	ReasoningItemID    string
+	ReasoningBuf       strings.Builder
+	ReasoningPartAdded bool
+	ReasoningIndex     int
+}
+
+var dataTag = []byte("data: ")
+
+func emitEvent(event string, payload string) string {
+	return fmt.Sprintf("event: %s\ndata: %s\n\n", event, payload)
+}
+
+// ConvertClaudeResponseToOpenAIResponses converts Claude SSE to OpenAI Responses SSE events.
+func ConvertClaudeResponseToOpenAIResponses(ctx context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, param *any) []string {
+	if *param == nil {
+		*param = &claudeToResponsesState{FuncArgsBuf: make(map[int]*strings.Builder), FuncNames: make(map[int]string), FuncCallIDs: make(map[int]string)}
+	}
+	st := (*param).(*claudeToResponsesState)
+
+	// Expect `data: {..}` from Claude clients
+	if !bytes.HasPrefix(rawJSON, dataTag) {
+		return []string{}
+	}
+	rawJSON = rawJSON[6:]
+	root := gjson.ParseBytes(rawJSON)
+	ev := root.Get("type").String()
+	var out []string
+
+	nextSeq := func() int { st.Seq++; return st.Seq }
+
+	switch ev {
+	case "message_start":
+		if msg := root.Get("message"); msg.Exists() {
+			st.ResponseID = msg.Get("id").String()
+			st.CreatedAt = time.Now().Unix()
+			// Reset per-message aggregation state
+			st.TextBuf.Reset()
+			st.ReasoningBuf.Reset()
+			st.ReasoningActive = false
+			st.InTextBlock = false
+			st.InFuncBlock = false
+			st.CurrentMsgID = ""
+			st.CurrentFCID = ""
+			st.ReasoningItemID = ""
+			st.ReasoningIndex = 0
+			st.ReasoningPartAdded = false
+			st.FuncArgsBuf = make(map[int]*strings.Builder)
+			st.FuncNames = make(map[int]string)
+			st.FuncCallIDs = make(map[int]string)
+			// response.created
+			created := `{"type":"response.created","sequence_number":0,"response":{"id":"","object":"response","created_at":0,"status":"in_progress","background":false,"error":null,"instructions":""}}`
+			created, _ = sjson.Set(created, "sequence_number", nextSeq())
+			created, _ = sjson.Set(created, "response.id", st.ResponseID)
+			created, _ = sjson.Set(created, "response.created_at", st.CreatedAt)
+			out = append(out, emitEvent("response.created", created))
+			// response.in_progress
+			inprog := `{"type":"response.in_progress","sequence_number":0,"response":{"id":"","object":"response","created_at":0,"status":"in_progress"}}`
+			inprog, _ = sjson.Set(inprog, "sequence_number", nextSeq())
+			inprog, _ = sjson.Set(inprog, "response.id", st.ResponseID)
+			inprog, _ = sjson.Set(inprog, "response.created_at", st.CreatedAt)
+			out = append(out, emitEvent("response.in_progress", inprog))
+		}
+	case "content_block_start":
+		cb := root.Get("content_block")
+		if !cb.Exists() {
+			return out
+		}
+		idx := int(root.Get("index").Int())
+		typ := cb.Get("type").String()
+		if typ == "text" {
+			// open message item + content part
+			st.InTextBlock = true
+			st.CurrentMsgID = fmt.Sprintf("msg_%s_0", st.ResponseID)
+			item := `{"type":"response.output_item.added","sequence_number":0,"output_index":0,"item":{"id":"","type":"message","status":"in_progress","content":[],"role":"assistant"}}`
+			item, _ = sjson.Set(item, "sequence_number", nextSeq())
+			item, _ = sjson.Set(item, "item.id", st.CurrentMsgID)
+			out = append(out, emitEvent("response.output_item.added", item))
+
+			part := `{"type":"response.content_part.added","sequence_number":0,"item_id":"","output_index":0,"content_index":0,"part":{"type":"output_text","annotations":[],"logprobs":[],"text":""}}`
+			part, _ = sjson.Set(part, "sequence_number", nextSeq())
+			part, _ = sjson.Set(part, "item_id", st.CurrentMsgID)
+			out = append(out, emitEvent("response.content_part.added", part))
+		} else if typ == "tool_use" {
+			st.InFuncBlock = true
+			st.CurrentFCID = cb.Get("id").String()
+			name := cb.Get("name").String()
+			item := `{"type":"response.output_item.added","sequence_number":0,"output_index":0,"item":{"id":"","type":"function_call","status":"in_progress","arguments":"","call_id":"","name":""}}`
+			item, _ = sjson.Set(item, "sequence_number", nextSeq())
+			item, _ = sjson.Set(item, "output_index", idx)
+			item, _ = sjson.Set(item, "item.id", fmt.Sprintf("fc_%s", st.CurrentFCID))
+			item, _ = sjson.Set(item, "item.call_id", st.CurrentFCID)
+			item, _ = sjson.Set(item, "item.name", name)
+			out = append(out, emitEvent("response.output_item.added", item))
+			if st.FuncArgsBuf[idx] == nil {
+				st.FuncArgsBuf[idx] = &strings.Builder{}
+			}
+			// record function metadata for aggregation
+			st.FuncCallIDs[idx] = st.CurrentFCID
+			st.FuncNames[idx] = name
+		} else if typ == "thinking" {
+			// start reasoning item
+			st.ReasoningActive = true
+			st.ReasoningIndex = idx
+			st.ReasoningBuf.Reset()
+			st.ReasoningItemID = fmt.Sprintf("rs_%s_%d", st.ResponseID, idx)
+			item := `{"type":"response.output_item.added","sequence_number":0,"output_index":0,"item":{"id":"","type":"reasoning","status":"in_progress","summary":[]}}`
+			item, _ = sjson.Set(item, "sequence_number", nextSeq())
+			item, _ = sjson.Set(item, "output_index", idx)
+			item, _ = sjson.Set(item, "item.id", st.ReasoningItemID)
+			out = append(out, emitEvent("response.output_item.added", item))
+			// add a summary part placeholder
+			part := `{"type":"response.reasoning_summary_part.added","sequence_number":0,"item_id":"","output_index":0,"summary_index":0,"part":{"type":"summary_text","text":""}}`
+			part, _ = sjson.Set(part, "sequence_number", nextSeq())
+			part, _ = sjson.Set(part, "item_id", st.ReasoningItemID)
+			part, _ = sjson.Set(part, "output_index", idx)
+			out = append(out, emitEvent("response.reasoning_summary_part.added", part))
+			st.ReasoningPartAdded = true
+		}
+	case "content_block_delta":
+		d := root.Get("delta")
+		if !d.Exists() {
+			return out
+		}
+		dt := d.Get("type").String()
+		if dt == "text_delta" {
+			if t := d.Get("text"); t.Exists() {
+				msg := `{"type":"response.output_text.delta","sequence_number":0,"item_id":"","output_index":0,"content_index":0,"delta":"","logprobs":[]}`
+				msg, _ = sjson.Set(msg, "sequence_number", nextSeq())
+				msg, _ = sjson.Set(msg, "item_id", st.CurrentMsgID)
+				msg, _ = sjson.Set(msg, "delta", t.String())
+				out = append(out, emitEvent("response.output_text.delta", msg))
+				// aggregate text for response.output
+				st.TextBuf.WriteString(t.String())
+			}
+		} else if dt == "input_json_delta" {
+			idx := int(root.Get("index").Int())
+			if pj := d.Get("partial_json"); pj.Exists() {
+				if st.FuncArgsBuf[idx] == nil {
+					st.FuncArgsBuf[idx] = &strings.Builder{}
+				}
+				st.FuncArgsBuf[idx].WriteString(pj.String())
+				msg := `{"type":"response.function_call_arguments.delta","sequence_number":0,"item_id":"","output_index":0,"delta":""}`
+				msg, _ = sjson.Set(msg, "sequence_number", nextSeq())
+				msg, _ = sjson.Set(msg, "item_id", fmt.Sprintf("fc_%s", st.CurrentFCID))
+				msg, _ = sjson.Set(msg, "output_index", idx)
+				msg, _ = sjson.Set(msg, "delta", pj.String())
+				out = append(out, emitEvent("response.function_call_arguments.delta", msg))
+			}
+		} else if dt == "thinking_delta" {
+			if st.ReasoningActive {
+				if t := d.Get("thinking"); t.Exists() {
+					st.ReasoningBuf.WriteString(t.String())
+					msg := `{"type":"response.reasoning_summary_text.delta","sequence_number":0,"item_id":"","output_index":0,"summary_index":0,"text":""}`
+					msg, _ = sjson.Set(msg, "sequence_number", nextSeq())
+					msg, _ = sjson.Set(msg, "item_id", st.ReasoningItemID)
+					msg, _ = sjson.Set(msg, "output_index", st.ReasoningIndex)
+					msg, _ = sjson.Set(msg, "text", t.String())
+					out = append(out, emitEvent("response.reasoning_summary_text.delta", msg))
+				}
+			}
+		}
+	case "content_block_stop":
+		idx := int(root.Get("index").Int())
+		if st.InTextBlock {
+			done := `{"type":"response.output_text.done","sequence_number":0,"item_id":"","output_index":0,"content_index":0,"text":"","logprobs":[]}`
+			done, _ = sjson.Set(done, "sequence_number", nextSeq())
+			done, _ = sjson.Set(done, "item_id", st.CurrentMsgID)
+			out = append(out, emitEvent("response.output_text.done", done))
+			partDone := `{"type":"response.content_part.done","sequence_number":0,"item_id":"","output_index":0,"content_index":0,"part":{"type":"output_text","annotations":[],"logprobs":[],"text":""}}`
+			partDone, _ = sjson.Set(partDone, "sequence_number", nextSeq())
+			partDone, _ = sjson.Set(partDone, "item_id", st.CurrentMsgID)
+			out = append(out, emitEvent("response.content_part.done", partDone))
+			final := `{"type":"response.output_item.done","sequence_number":0,"output_index":0,"item":{"id":"","type":"message","status":"completed","content":[{"type":"output_text","text":""}],"role":"assistant"}}`
+			final, _ = sjson.Set(final, "sequence_number", nextSeq())
+			final, _ = sjson.Set(final, "item.id", st.CurrentMsgID)
+			out = append(out, emitEvent("response.output_item.done", final))
+			st.InTextBlock = false
+		} else if st.InFuncBlock {
+			args := "{}"
+			if buf := st.FuncArgsBuf[idx]; buf != nil {
+				if buf.Len() > 0 {
+					args = buf.String()
+				}
+			}
+			fcDone := `{"type":"response.function_call_arguments.done","sequence_number":0,"item_id":"","output_index":0,"arguments":""}`
+			fcDone, _ = sjson.Set(fcDone, "sequence_number", nextSeq())
+			fcDone, _ = sjson.Set(fcDone, "item_id", fmt.Sprintf("fc_%s", st.CurrentFCID))
+			fcDone, _ = sjson.Set(fcDone, "output_index", idx)
+			fcDone, _ = sjson.Set(fcDone, "arguments", args)
+			out = append(out, emitEvent("response.function_call_arguments.done", fcDone))
+			itemDone := `{"type":"response.output_item.done","sequence_number":0,"output_index":0,"item":{"id":"","type":"function_call","status":"completed","arguments":"","call_id":"","name":""}}`
+			itemDone, _ = sjson.Set(itemDone, "sequence_number", nextSeq())
+			itemDone, _ = sjson.Set(itemDone, "output_index", idx)
+			itemDone, _ = sjson.Set(itemDone, "item.id", fmt.Sprintf("fc_%s", st.CurrentFCID))
+			itemDone, _ = sjson.Set(itemDone, "item.arguments", args)
+			itemDone, _ = sjson.Set(itemDone, "item.call_id", st.CurrentFCID)
+			out = append(out, emitEvent("response.output_item.done", itemDone))
+			st.InFuncBlock = false
+		} else if st.ReasoningActive {
+			// close reasoning
+			full := st.ReasoningBuf.String()
+			textDone := `{"type":"response.reasoning_summary_text.done","sequence_number":0,"item_id":"","output_index":0,"summary_index":0,"text":""}`
+			textDone, _ = sjson.Set(textDone, "sequence_number", nextSeq())
+			textDone, _ = sjson.Set(textDone, "item_id", st.ReasoningItemID)
+			textDone, _ = sjson.Set(textDone, "output_index", st.ReasoningIndex)
+			textDone, _ = sjson.Set(textDone, "text", full)
+			out = append(out, emitEvent("response.reasoning_summary_text.done", textDone))
+			partDone := `{"type":"response.reasoning_summary_part.done","sequence_number":0,"item_id":"","output_index":0,"summary_index":0,"part":{"type":"summary_text","text":""}}`
+			partDone, _ = sjson.Set(partDone, "sequence_number", nextSeq())
+			partDone, _ = sjson.Set(partDone, "item_id", st.ReasoningItemID)
+			partDone, _ = sjson.Set(partDone, "output_index", st.ReasoningIndex)
+			partDone, _ = sjson.Set(partDone, "part.text", full)
+			out = append(out, emitEvent("response.reasoning_summary_part.done", partDone))
+			st.ReasoningActive = false
+			st.ReasoningPartAdded = false
+		}
+	case "message_stop":
+		completed := `{"type":"response.completed","sequence_number":0,"response":{"id":"","object":"response","created_at":0,"status":"completed","background":false,"error":null}}`
+		completed, _ = sjson.Set(completed, "sequence_number", nextSeq())
+		completed, _ = sjson.Set(completed, "response.id", st.ResponseID)
+		completed, _ = sjson.Set(completed, "response.created_at", st.CreatedAt)
+		// Inject original request fields into response as per docs/response.completed.json
+
+		if requestRawJSON != nil {
+			req := gjson.ParseBytes(requestRawJSON)
+			if v := req.Get("instructions"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.instructions", v.String())
+			}
+			if v := req.Get("max_output_tokens"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.max_output_tokens", v.Int())
+			}
+			if v := req.Get("max_tool_calls"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.max_tool_calls", v.Int())
+			}
+			if v := req.Get("model"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.model", v.String())
+			}
+			if v := req.Get("parallel_tool_calls"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.parallel_tool_calls", v.Bool())
+			}
+			if v := req.Get("previous_response_id"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.previous_response_id", v.String())
+			}
+			if v := req.Get("prompt_cache_key"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.prompt_cache_key", v.String())
+			}
+			if v := req.Get("reasoning"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.reasoning", v.Value())
+			}
+			if v := req.Get("safety_identifier"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.safety_identifier", v.String())
+			}
+			if v := req.Get("service_tier"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.service_tier", v.String())
+			}
+			if v := req.Get("store"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.store", v.Bool())
+			}
+			if v := req.Get("temperature"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.temperature", v.Float())
+			}
+			if v := req.Get("text"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.text", v.Value())
+			}
+			if v := req.Get("tool_choice"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.tool_choice", v.Value())
+			}
+			if v := req.Get("tools"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.tools", v.Value())
+			}
+			if v := req.Get("top_logprobs"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.top_logprobs", v.Int())
+			}
+			if v := req.Get("top_p"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.top_p", v.Float())
+			}
+			if v := req.Get("truncation"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.truncation", v.String())
+			}
+			if v := req.Get("user"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.user", v.Value())
+			}
+			if v := req.Get("metadata"); v.Exists() {
+				completed, _ = sjson.Set(completed, "response.metadata", v.Value())
+			}
+		}
+
+		// Build response.output from aggregated state
+		var outputs []interface{}
+		// reasoning item (if any)
+		if st.ReasoningBuf.Len() > 0 || st.ReasoningPartAdded {
+			r := map[string]interface{}{
+				"id":      st.ReasoningItemID,
+				"type":    "reasoning",
+				"summary": []interface{}{map[string]interface{}{"type": "summary_text", "text": st.ReasoningBuf.String()}},
+			}
+			outputs = append(outputs, r)
+		}
+		// assistant message item (if any text)
+		if st.TextBuf.Len() > 0 || st.InTextBlock || st.CurrentMsgID != "" {
+			m := map[string]interface{}{
+				"id":     st.CurrentMsgID,
+				"type":   "message",
+				"status": "completed",
+				"content": []interface{}{map[string]interface{}{
+					"type":        "output_text",
+					"annotations": []interface{}{},
+					"logprobs":    []interface{}{},
+					"text":        st.TextBuf.String(),
+				}},
+				"role": "assistant",
+			}
+			outputs = append(outputs, m)
+		}
+		// function_call items (in ascending index order for determinism)
+		if len(st.FuncArgsBuf) > 0 {
+			// collect indices
+			idxs := make([]int, 0, len(st.FuncArgsBuf))
+			for idx := range st.FuncArgsBuf {
+				idxs = append(idxs, idx)
+			}
+			// simple sort (small N), avoid adding new imports
+			for i := 0; i < len(idxs); i++ {
+				for j := i + 1; j < len(idxs); j++ {
+					if idxs[j] < idxs[i] {
+						idxs[i], idxs[j] = idxs[j], idxs[i]
+					}
+				}
+			}
+			for _, idx := range idxs {
+				args := ""
+				if b := st.FuncArgsBuf[idx]; b != nil {
+					args = b.String()
+				}
+				callID := st.FuncCallIDs[idx]
+				name := st.FuncNames[idx]
+				if callID == "" && st.CurrentFCID != "" {
+					callID = st.CurrentFCID
+				}
+				item := map[string]interface{}{
+					"id":        fmt.Sprintf("fc_%s", callID),
+					"type":      "function_call",
+					"status":    "completed",
+					"arguments": args,
+					"call_id":   callID,
+					"name":      name,
+				}
+				outputs = append(outputs, item)
+			}
+		}
+		if len(outputs) > 0 {
+			completed, _ = sjson.Set(completed, "response.output", outputs)
+		}
+		out = append(out, emitEvent("response.completed", completed))
+	}
+
+	return out
+}
+
+// ConvertClaudeResponseToOpenAIResponsesNonStream aggregates Claude SSE into a single OpenAI Responses JSON.
+func ConvertClaudeResponseToOpenAIResponsesNonStream(_ context.Context, _ string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, _ *any) string {
+	// Aggregate Claude SSE lines into a single OpenAI Responses JSON (non-stream)
+	// We follow the same aggregation logic as the streaming variant but produce
+	// one final object matching docs/out.json structure.
+
+	// Collect SSE data: lines start with "data: "; ignore others
+	var chunks [][]byte
+	{
+		// Use a simple scanner to iterate through raw bytes
+		// Note: extremely large responses may require increasing the buffer
+		scanner := bufio.NewScanner(bytes.NewReader(rawJSON))
+		buf := make([]byte, 10240*1024)
+		scanner.Buffer(buf, 10240*1024)
+		for scanner.Scan() {
+			line := scanner.Bytes()
+			if !bytes.HasPrefix(line, dataTag) {
+				continue
+			}
+			chunks = append(chunks, line[len(dataTag):])
+		}
+	}
+
+	// Base OpenAI Responses (non-stream) object
+	out := `{"id":"","object":"response","created_at":0,"status":"completed","background":false,"error":null,"incomplete_details":null,"output":[],"usage":{"input_tokens":0,"input_tokens_details":{"cached_tokens":0},"output_tokens":0,"output_tokens_details":{},"total_tokens":0}}`
+
+	// Aggregation state
+	var (
+		responseID      string
+		createdAt       int64
+		currentMsgID    string
+		currentFCID     string
+		textBuf         strings.Builder
+		reasoningBuf    strings.Builder
+		reasoningActive bool
+		reasoningItemID string
+		inputTokens     int64
+		outputTokens    int64
+	)
+
+	// Per-index tool call aggregation
+	type toolState struct {
+		id   string
+		name string
+		args strings.Builder
+	}
+	toolCalls := make(map[int]*toolState)
+
+	// Walk through SSE chunks to fill state
+	for _, ch := range chunks {
+		root := gjson.ParseBytes(ch)
+		ev := root.Get("type").String()
+
+		switch ev {
+		case "message_start":
+			if msg := root.Get("message"); msg.Exists() {
+				responseID = msg.Get("id").String()
+				createdAt = time.Now().Unix()
+				if usage := msg.Get("usage"); usage.Exists() {
+					inputTokens = usage.Get("input_tokens").Int()
+				}
+			}
+
+		case "content_block_start":
+			cb := root.Get("content_block")
+			if !cb.Exists() {
+				continue
+			}
+			idx := int(root.Get("index").Int())
+			typ := cb.Get("type").String()
+			switch typ {
+			case "text":
+				currentMsgID = "msg_" + responseID + "_0"
+			case "tool_use":
+				currentFCID = cb.Get("id").String()
+				name := cb.Get("name").String()
+				if toolCalls[idx] == nil {
+					toolCalls[idx] = &toolState{id: currentFCID, name: name}
+				} else {
+					toolCalls[idx].id = currentFCID
+					toolCalls[idx].name = name
+				}
+			case "thinking":
+				reasoningActive = true
+				reasoningItemID = fmt.Sprintf("rs_%s_%d", responseID, idx)
+			}
+
+		case "content_block_delta":
+			d := root.Get("delta")
+			if !d.Exists() {
+				continue
+			}
+			dt := d.Get("type").String()
+			switch dt {
+			case "text_delta":
+				if t := d.Get("text"); t.Exists() {
+					textBuf.WriteString(t.String())
+				}
+			case "input_json_delta":
+				if pj := d.Get("partial_json"); pj.Exists() {
+					idx := int(root.Get("index").Int())
+					if toolCalls[idx] == nil {
+						toolCalls[idx] = &toolState{}
+					}
+					toolCalls[idx].args.WriteString(pj.String())
+				}
+			case "thinking_delta":
+				if reasoningActive {
+					if t := d.Get("thinking"); t.Exists() {
+						reasoningBuf.WriteString(t.String())
+					}
+				}
+			}
+
+		case "content_block_stop":
+			// Nothing special to finalize for non-stream aggregation
+			_ = root
+
+		case "message_delta":
+			if usage := root.Get("usage"); usage.Exists() {
+				outputTokens = usage.Get("output_tokens").Int()
+			}
+		}
+	}
+
+	// Populate base fields
+	out, _ = sjson.Set(out, "id", responseID)
+	out, _ = sjson.Set(out, "created_at", createdAt)
+
+	// Inject request echo fields as top-level (similar to streaming variant)
+	if requestRawJSON != nil {
+		req := gjson.ParseBytes(requestRawJSON)
+		if v := req.Get("instructions"); v.Exists() {
+			out, _ = sjson.Set(out, "instructions", v.String())
+		}
+		if v := req.Get("max_output_tokens"); v.Exists() {
+			out, _ = sjson.Set(out, "max_output_tokens", v.Int())
+		}
+		if v := req.Get("max_tool_calls"); v.Exists() {
+			out, _ = sjson.Set(out, "max_tool_calls", v.Int())
+		}
+		if v := req.Get("model"); v.Exists() {
+			out, _ = sjson.Set(out, "model", v.String())
+		}
+		if v := req.Get("parallel_tool_calls"); v.Exists() {
+			out, _ = sjson.Set(out, "parallel_tool_calls", v.Bool())
+		}
+		if v := req.Get("previous_response_id"); v.Exists() {
+			out, _ = sjson.Set(out, "previous_response_id", v.String())
+		}
+		if v := req.Get("prompt_cache_key"); v.Exists() {
+			out, _ = sjson.Set(out, "prompt_cache_key", v.String())
+		}
+		if v := req.Get("reasoning"); v.Exists() {
+			out, _ = sjson.Set(out, "reasoning", v.Value())
+		}
+		if v := req.Get("safety_identifier"); v.Exists() {
+			out, _ = sjson.Set(out, "safety_identifier", v.String())
+		}
+		if v := req.Get("service_tier"); v.Exists() {
+			out, _ = sjson.Set(out, "service_tier", v.String())
+		}
+		if v := req.Get("store"); v.Exists() {
+			out, _ = sjson.Set(out, "store", v.Bool())
+		}
+		if v := req.Get("temperature"); v.Exists() {
+			out, _ = sjson.Set(out, "temperature", v.Float())
+		}
+		if v := req.Get("text"); v.Exists() {
+			out, _ = sjson.Set(out, "text", v.Value())
+		}
+		if v := req.Get("tool_choice"); v.Exists() {
+			out, _ = sjson.Set(out, "tool_choice", v.Value())
+		}
+		if v := req.Get("tools"); v.Exists() {
+			out, _ = sjson.Set(out, "tools", v.Value())
+		}
+		if v := req.Get("top_logprobs"); v.Exists() {
+			out, _ = sjson.Set(out, "top_logprobs", v.Int())
+		}
+		if v := req.Get("top_p"); v.Exists() {
+			out, _ = sjson.Set(out, "top_p", v.Float())
+		}
+		if v := req.Get("truncation"); v.Exists() {
+			out, _ = sjson.Set(out, "truncation", v.String())
+		}
+		if v := req.Get("user"); v.Exists() {
+			out, _ = sjson.Set(out, "user", v.Value())
+		}
+		if v := req.Get("metadata"); v.Exists() {
+			out, _ = sjson.Set(out, "metadata", v.Value())
+		}
+	}
+
+	// Build output array
+	var outputs []interface{}
+	if reasoningBuf.Len() > 0 {
+		outputs = append(outputs, map[string]interface{}{
+			"id":      reasoningItemID,
+			"type":    "reasoning",
+			"summary": []interface{}{map[string]interface{}{"type": "summary_text", "text": reasoningBuf.String()}},
+		})
+	}
+	if currentMsgID != "" || textBuf.Len() > 0 {
+		outputs = append(outputs, map[string]interface{}{
+			"id":     currentMsgID,
+			"type":   "message",
+			"status": "completed",
+			"content": []interface{}{map[string]interface{}{
+				"type":        "output_text",
+				"annotations": []interface{}{},
+				"logprobs":    []interface{}{},
+				"text":        textBuf.String(),
+			}},
+			"role": "assistant",
+		})
+	}
+	if len(toolCalls) > 0 {
+		// Preserve index order
+		idxs := make([]int, 0, len(toolCalls))
+		for i := range toolCalls {
+			idxs = append(idxs, i)
+		}
+		for i := 0; i < len(idxs); i++ {
+			for j := i + 1; j < len(idxs); j++ {
+				if idxs[j] < idxs[i] {
+					idxs[i], idxs[j] = idxs[j], idxs[i]
+				}
+			}
+		}
+		for _, i := range idxs {
+			st := toolCalls[i]
+			args := st.args.String()
+			if args == "" {
+				args = "{}"
+			}
+			outputs = append(outputs, map[string]interface{}{
+				"id":        fmt.Sprintf("fc_%s", st.id),
+				"type":      "function_call",
+				"status":    "completed",
+				"arguments": args,
+				"call_id":   st.id,
+				"name":      st.name,
+			})
+		}
+	}
+	if len(outputs) > 0 {
+		out, _ = sjson.Set(out, "output", outputs)
+	}
+
+	// Usage
+	total := inputTokens + outputTokens
+	out, _ = sjson.Set(out, "usage.input_tokens", inputTokens)
+	out, _ = sjson.Set(out, "usage.output_tokens", outputTokens)
+	out, _ = sjson.Set(out, "usage.total_tokens", total)
+	if reasoningBuf.Len() > 0 {
+		// Rough estimate similar to chat completions
+		reasoningTokens := int64(len(reasoningBuf.String()) / 4)
+		if reasoningTokens > 0 {
+			out, _ = sjson.Set(out, "usage.output_tokens_details.reasoning_tokens", reasoningTokens)
+		}
+	}
+
+	return out
+}

internal/translator/claude/openai/responses/init.go

@@ -0,0 +1,19 @@
+package responses
+
+import (
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/translator/translator"
+)
+
+func init() {
+	translator.Register(
+		OPENAI_RESPONSE,
+		CLAUDE,
+		ConvertOpenAIResponsesRequestToClaude,
+		interfaces.TranslateResponse{
+			Stream:    ConvertClaudeResponseToOpenAIResponses,
+			NonStream: ConvertClaudeResponseToOpenAIResponsesNonStream,
+		},
+	)
+}

internal/translator/codex/claude/codex_claude_request.go

@@ -6,9 +6,12 @@
 package claude
 
 import (
+	"bytes"
 	"fmt"
+	"strconv"
+	"strings"
 
-	"github.com/luispater/CLIProxyAPI/internal/misc"
+	"github.com/luispater/CLIProxyAPI/v5/internal/misc"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
 )
@@ -31,10 +34,12 @@ import (
 //
 // Returns:
 //   - []byte: The transformed request data in internal client format
-func ConvertClaudeRequestToCodex(modelName string, rawJSON []byte, _ bool) []byte {
+func ConvertClaudeRequestToCodex(modelName string, inputRawJSON []byte, _ bool) []byte {
+	rawJSON := bytes.Clone(inputRawJSON)
+
 	template := `{"model":"","instructions":"","input":[]}`
 
-	instructions := misc.CodexInstructions
+	instructions := misc.CodexInstructions(modelName)
 	template, _ = sjson.SetRaw(template, "instructions", instructions)
 
 	rootResult := gjson.ParseBytes(rawJSON)
@@ -91,7 +96,17 @@ func ConvertClaudeRequestToCodex(modelName string, rawJSON []byte, _ bool) []byt
 						// Handle tool use content by creating function call message.
 						functionCallMessage := `{"type":"function_call"}`
 						functionCallMessage, _ = sjson.Set(functionCallMessage, "call_id", messageContentResult.Get("id").String())
-						functionCallMessage, _ = sjson.Set(functionCallMessage, "name", messageContentResult.Get("name").String())
+						{
+							// Shorten tool name if needed based on declared tools
+							name := messageContentResult.Get("name").String()
+							toolMap := buildReverseMapFromClaudeOriginalToShort(rawJSON)
+							if short, ok := toolMap[name]; ok {
+								name = short
+							} else {
+								name = shortenNameIfNeeded(name)
+							}
+							functionCallMessage, _ = sjson.Set(functionCallMessage, "name", name)
+						}
 						functionCallMessage, _ = sjson.Set(functionCallMessage, "arguments", messageContentResult.Get("input").Raw)
 						template, _ = sjson.SetRaw(template, "input.-1", functionCallMessage)
 					} else if contentType == "tool_result" {
@@ -127,10 +142,29 @@ func ConvertClaudeRequestToCodex(modelName string, rawJSON []byte, _ bool) []byt
 		template, _ = sjson.SetRaw(template, "tools", `[]`)
 		template, _ = sjson.Set(template, "tool_choice", `auto`)
 		toolResults := toolsResult.Array()
+		// Build short name map from declared tools
+		var names []string
+		for i := 0; i < len(toolResults); i++ {
+			n := toolResults[i].Get("name").String()
+			if n != "" {
+				names = append(names, n)
+			}
+		}
+		shortMap := buildShortNameMap(names)
 		for i := 0; i < len(toolResults); i++ {
 			toolResult := toolResults[i]
 			tool := toolResult.Raw
 			tool, _ = sjson.Set(tool, "type", "function")
+			// Apply shortened name if needed
+			if v := toolResult.Get("name"); v.Exists() {
+				name := v.String()
+				if short, ok := shortMap[name]; ok {
+					name = short
+				} else {
+					name = shortenNameIfNeeded(name)
+				}
+				tool, _ = sjson.Set(tool, "name", name)
+			}
 			tool, _ = sjson.SetRaw(tool, "parameters", toolResult.Get("input_schema").Raw)
 			tool, _ = sjson.Delete(tool, "input_schema")
 			tool, _ = sjson.Delete(tool, "parameters.$schema")
@@ -167,3 +201,97 @@ func ConvertClaudeRequestToCodex(modelName string, rawJSON []byte, _ bool) []byt
 
 	return []byte(template)
 }
+
+// shortenNameIfNeeded applies a simple shortening rule for a single name.
+func shortenNameIfNeeded(name string) string {
+	const limit = 64
+	if len(name) <= limit {
+		return name
+	}
+	if strings.HasPrefix(name, "mcp__") {
+		idx := strings.LastIndex(name, "__")
+		if idx > 0 {
+			cand := "mcp__" + name[idx+2:]
+			if len(cand) > limit {
+				return cand[:limit]
+			}
+			return cand
+		}
+	}
+	return name[:limit]
+}
+
+// buildShortNameMap ensures uniqueness of shortened names within a request.
+func buildShortNameMap(names []string) map[string]string {
+	const limit = 64
+	used := map[string]struct{}{}
+	m := map[string]string{}
+
+	baseCandidate := func(n string) string {
+		if len(n) <= limit {
+			return n
+		}
+		if strings.HasPrefix(n, "mcp__") {
+			idx := strings.LastIndex(n, "__")
+			if idx > 0 {
+				cand := "mcp__" + n[idx+2:]
+				if len(cand) > limit {
+					cand = cand[:limit]
+				}
+				return cand
+			}
+		}
+		return n[:limit]
+	}
+
+	makeUnique := func(cand string) string {
+		if _, ok := used[cand]; !ok {
+			return cand
+		}
+		base := cand
+		for i := 1; ; i++ {
+			suffix := "~" + strconv.Itoa(i)
+			allowed := limit - len(suffix)
+			if allowed < 0 {
+				allowed = 0
+			}
+			tmp := base
+			if len(tmp) > allowed {
+				tmp = tmp[:allowed]
+			}
+			tmp = tmp + suffix
+			if _, ok := used[tmp]; !ok {
+				return tmp
+			}
+		}
+	}
+
+	for _, n := range names {
+		cand := baseCandidate(n)
+		uniq := makeUnique(cand)
+		used[uniq] = struct{}{}
+		m[n] = uniq
+	}
+	return m
+}
+
+// buildReverseMapFromClaudeOriginalToShort builds original->short map, used to map tool_use names to short.
+func buildReverseMapFromClaudeOriginalToShort(original []byte) map[string]string {
+	tools := gjson.GetBytes(original, "tools")
+	m := map[string]string{}
+	if !tools.IsArray() {
+		return m
+	}
+	var names []string
+	arr := tools.Array()
+	for i := 0; i < len(arr); i++ {
+		n := arr[i].Get("name").String()
+		if n != "" {
+			names = append(names, n)
+		}
+	}
+	if len(names) > 0 {
+		m = buildShortNameMap(names)
+	}
+	return m
+}

internal/translator/codex/claude/codex_claude_response.go

@@ -35,7 +35,7 @@ var (
 //
 // Returns:
 //   - []string: A slice of strings, each containing a Claude Code-compatible JSON response
-func ConvertCodexResponseToClaude(_ context.Context, _ string, rawJSON []byte, param *any) []string {
+func ConvertCodexResponseToClaude(_ context.Context, _ string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, param *any) []string {
 	if *param == nil {
 		hasToolCall := false
 		*param = &hasToolCall
@@ -122,7 +122,15 @@ func ConvertCodexResponseToClaude(_ context.Context, _ string, rawJSON []byte, p
 			template = `{"type":"content_block_start","index":0,"content_block":{"type":"tool_use","id":"","name":"","input":{}}}`
 			template, _ = sjson.Set(template, "index", rootResult.Get("output_index").Int())
 			template, _ = sjson.Set(template, "content_block.id", itemResult.Get("call_id").String())
-			template, _ = sjson.Set(template, "content_block.name", itemResult.Get("name").String())
+			{
+				// Restore original tool name if shortened
+				name := itemResult.Get("name").String()
+				rev := buildReverseMapFromClaudeOriginalShortToOriginal(originalRequestRawJSON)
+				if orig, ok := rev[name]; ok {
+					name = orig
+				}
+				template, _ = sjson.Set(template, "content_block.name", name)
+			}
 
 			output = "event: content_block_start\n"
 			output += fmt.Sprintf("data: %s\n\n", template)
@@ -168,6 +176,30 @@ func ConvertCodexResponseToClaude(_ context.Context, _ string, rawJSON []byte, p
 //
 // Returns:
 //   - string: A Claude Code-compatible JSON response containing all message content and metadata
-func ConvertCodexResponseToClaudeNonStream(_ context.Context, _ string, _ []byte, _ *any) string {
+func ConvertCodexResponseToClaudeNonStream(_ context.Context, _ string, originalRequestRawJSON, requestRawJSON, _ []byte, _ *any) string {
 	return ""
 }
+
+// buildReverseMapFromClaudeOriginalShortToOriginal builds a map[short]original from original Claude request tools.
+func buildReverseMapFromClaudeOriginalShortToOriginal(original []byte) map[string]string {
+	tools := gjson.GetBytes(original, "tools")
+	rev := map[string]string{}
+	if !tools.IsArray() {
+		return rev
+	}
+	var names []string
+	arr := tools.Array()
+	for i := 0; i < len(arr); i++ {
+		n := arr[i].Get("name").String()
+		if n != "" {
+			names = append(names, n)
+		}
+	}
+	if len(names) > 0 {
+		m := buildShortNameMap(names)
+		for orig, short := range m {
+			rev[short] = orig
+		}
+	}
+	return rev
+}

internal/translator/codex/claude/init.go

@@ -1,9 +1,9 @@
 package claude
 
 import (
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/translator/translator"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/translator/translator"
 )
 
 func init() {

internal/translator/codex/gemini-cli/codex_gemini-cli_request.go

@@ -6,7 +6,9 @@
 package geminiCLI
 
 import (
-	. "github.com/luispater/CLIProxyAPI/internal/translator/codex/gemini"
+	"bytes"
+
+	. "github.com/luispater/CLIProxyAPI/v5/internal/translator/codex/gemini"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
 )
@@ -27,7 +29,9 @@ import (
 //
 // Returns:
 //   - []byte: The transformed request data in Codex API format
-func ConvertGeminiCLIRequestToCodex(modelName string, rawJSON []byte, stream bool) []byte {
+func ConvertGeminiCLIRequestToCodex(modelName string, inputRawJSON []byte, stream bool) []byte {
+	rawJSON := bytes.Clone(inputRawJSON)
+
 	rawJSON = []byte(gjson.GetBytes(rawJSON, "request").Raw)
 	rawJSON, _ = sjson.SetBytes(rawJSON, "model", modelName)
 	if gjson.GetBytes(rawJSON, "systemInstruction").Exists() {

internal/translator/codex/gemini-cli/codex_gemini-cli_response.go

@@ -7,7 +7,7 @@ package geminiCLI
 import (
 	"context"
 
-	. "github.com/luispater/CLIProxyAPI/internal/translator/codex/gemini"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/translator/codex/gemini"
 	"github.com/tidwall/sjson"
 )
 
@@ -24,8 +24,8 @@ import (
 //
 // Returns:
 //   - []string: A slice of strings, each containing a Gemini-compatible JSON response wrapped in a response object
-func ConvertCodexResponseToGeminiCLI(ctx context.Context, modelName string, rawJSON []byte, param *any) []string {
-	outputs := ConvertCodexResponseToGemini(ctx, modelName, rawJSON, param)
+func ConvertCodexResponseToGeminiCLI(ctx context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, param *any) []string {
+	outputs := ConvertCodexResponseToGemini(ctx, modelName, originalRequestRawJSON, requestRawJSON, rawJSON, param)
 	newOutputs := make([]string, 0)
 	for i := 0; i < len(outputs); i++ {
 		json := `{"response": {}}`
@@ -47,9 +47,9 @@ func ConvertCodexResponseToGeminiCLI(ctx context.Context, modelName string, rawJ
 //
 // Returns:
 //   - string: A Gemini-compatible JSON response wrapped in a response object
-func ConvertCodexResponseToGeminiCLINonStream(ctx context.Context, modelName string, rawJSON []byte, param *any) string {
+func ConvertCodexResponseToGeminiCLINonStream(ctx context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, param *any) string {
 	// log.Debug(string(rawJSON))
-	strJSON := ConvertCodexResponseToGeminiNonStream(ctx, modelName, rawJSON, param)
+	strJSON := ConvertCodexResponseToGeminiNonStream(ctx, modelName, originalRequestRawJSON, requestRawJSON, rawJSON, param)
 	json := `{"response": {}}`
 	strJSON, _ = sjson.SetRaw(json, "response", strJSON)
 	return strJSON

internal/translator/codex/gemini-cli/init.go

@@ -1,9 +1,9 @@
 package geminiCLI
 
 import (
-	. "github.com/luispater/CLIProxyAPI/internal/constant"
-	"github.com/luispater/CLIProxyAPI/internal/interfaces"
-	"github.com/luispater/CLIProxyAPI/internal/translator/translator"
+	. "github.com/luispater/CLIProxyAPI/v5/internal/constant"
+	"github.com/luispater/CLIProxyAPI/v5/internal/interfaces"
+	"github.com/luispater/CLIProxyAPI/v5/internal/translator/translator"
 )
 
 func init() {

internal/translator/codex/gemini/codex_gemini_request.go

@@ -6,13 +6,15 @@
 package gemini
 
 import (
+	"bytes"
 	"crypto/rand"
 	"fmt"
 	"math/big"
+	"strconv"
 	"strings"
 
-	"github.com/luispater/CLIProxyAPI/internal/misc"
-	"github.com/luispater/CLIProxyAPI/internal/util"
+	"github.com/luispater/CLIProxyAPI/v5/internal/misc"
+	"github.com/luispater/CLIProxyAPI/v5/internal/util"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
 )
@@ -34,16 +36,38 @@ import (
 //
 // Returns:
 //   - []byte: The transformed request data in Codex API format
-func ConvertGeminiRequestToCodex(modelName string, rawJSON []byte, _ bool) []byte {
+func ConvertGeminiRequestToCodex(modelName string, inputRawJSON []byte, _ bool) []byte {
+	rawJSON := bytes.Clone(inputRawJSON)
 	// Base template
 	out := `{"model":"","instructions":"","input":[]}`
 
 	// Inject standard Codex instructions
-	instructions := misc.CodexInstructions
+	instructions := misc.CodexInstructions(modelName)
 	out, _ = sjson.SetRaw(out, "instructions", instructions)
 
 	root := gjson.ParseBytes(rawJSON)
 
+	// Pre-compute tool name shortening map from declared functionDeclarations
+	shortMap := map[string]string{}
+	if tools := root.Get("tools"); tools.IsArray() {
+		var names []string
+		tarr := tools.Array()
+		for i := 0; i < len(tarr); i++ {
+			fns := tarr[i].Get("functionDeclarations")
+			if !fns.IsArray() {
+				continue
+			}
+			for _, fn := range fns.Array() {
+				if v := fn.Get("name"); v.Exists() {
+					names = append(names, v.String())
+				}
+			}
+		}
+		if len(names) > 0 {
+			shortMap = buildShortNameMap(names)
+		}
+	}
+
 	// helper for generating paired call IDs in the form: call_<alphanum>
 	// Gemini uses sequential pairing across possibly multiple in-flight
 	// functionCalls, so we keep a FIFO queue of generated call IDs and
@@ -122,7 +146,13 @@ func ConvertGeminiRequestToCodex(modelName string, rawJSON []byte, _ bool) []byt
 				if fc := p.Get("functionCall"); fc.Exists() {
 					fn := `{"type":"function_call"}`
 					if name := fc.Get("name"); name.Exists() {
-						fn, _ = sjson.Set(fn, "name", name.String())
+						n := name.String()
+						if short, ok := shortMap[n]; ok {
+							n = short
+						} else {
+							n = shortenNameIfNeeded(n)
+						}
+						fn, _ = sjson.Set(fn, "name", n)
 					}
 					if args := fc.Get("args"); args.Exists() {
 						fn, _ = sjson.Set(fn, "arguments", args.Raw)
@@ -183,7 +213,13 @@ func ConvertGeminiRequestToCodex(modelName string, rawJSON []byte, _ bool) []byt
 				tool := `{}`
 				tool, _ = sjson.Set(tool, "type", "function")
 				if v := fn.Get("name"); v.Exists() {
-					tool, _ = sjson.Set(tool, "name", v.String())
+					name := v.String()
+					if short, ok := shortMap[name]; ok {
+						name = short
+					} else {
+						name = shortenNameIfNeeded(name)
+					}
+					tool, _ = sjson.Set(tool, "name", name)
 				}
 				if v := fn.Get("description"); v.Exists() {
 					tool, _ = sjson.Set(tool, "description", v.String())
@@ -225,3 +261,76 @@ func ConvertGeminiRequestToCodex(modelName string, rawJSON []byte, _ bool) []byt
 
 	return []byte(out)
 }
+
+// shortenNameIfNeeded applies the simple shortening rule for a single name.
+func shortenNameIfNeeded(name string) string {
+	const limit = 64
+	if len(name) <= limit {
+		return name
+	}
+	if strings.HasPrefix(name, "mcp__") {
+		idx := strings.LastIndex(name, "__")
+		if idx > 0 {
+			cand := "mcp__" + name[idx+2:]
+			if len(cand) > limit {
+				return cand[:limit]
+			}
+			return cand
+		}
+	}
+	return name[:limit]
+}
+
+// buildShortNameMap ensures uniqueness of shortened names within a request.
+func buildShortNameMap(names []string) map[string]string {
+	const limit = 64
+	used := map[string]struct{}{}
+	m := map[string]string{}
+
+	baseCandidate := func(n string) string {
+		if len(n) <= limit {
+			return n
+		}
+		if strings.HasPrefix(n, "mcp__") {
+			idx := strings.LastIndex(n, "__")
+			if idx > 0 {
+				cand := "mcp__" + n[idx+2:]
+				if len(cand) > limit {
+					cand = cand[:limit]
+				}
+				return cand
+			}
+		}
+		return n[:limit]
+	}
+
+	makeUnique := func(cand string) string {
+		if _, ok := used[cand]; !ok {
+			return cand
+		}
+		base := cand
+		for i := 1; ; i++ {
+			suffix := "~" + strconv.Itoa(i)
+			allowed := limit - len(suffix)
+			if allowed < 0 {
+				allowed = 0
+			}
+			tmp := base
+			if len(tmp) > allowed {
+				tmp = tmp[:allowed]
+			}
+			tmp = tmp + suffix
+			if _, ok := used[tmp]; !ok {
+				return tmp
+			}
+		}
+	}
+
+	for _, n := range names {
+		cand := baseCandidate(n)
+		uniq := makeUnique(cand)
+		used[uniq] = struct{}{}
+		m[n] = uniq
+	}
+	return m
+}

internal/translator/codex/gemini/codex_gemini_response.go

@@ -40,7 +40,7 @@ type ConvertCodexResponseToGeminiParams struct {
 //
 // Returns:
 //   - []string: A slice of strings, each containing a Gemini-compatible JSON response
-func ConvertCodexResponseToGemini(_ context.Context, modelName string, rawJSON []byte, param *any) []string {
+func ConvertCodexResponseToGemini(_ context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, param *any) []string {
 	if *param == nil {
 		*param = &ConvertCodexResponseToGeminiParams{
 			Model:             modelName,
@@ -80,7 +80,15 @@ func ConvertCodexResponseToGemini(_ context.Context, modelName string, rawJSON [
 		if itemType == "function_call" {
 			// Create function call part
 			functionCall := `{"functionCall":{"name":"","args":{}}}`
-			functionCall, _ = sjson.Set(functionCall, "functionCall.name", itemResult.Get("name").String())
+			{
+				// Restore original tool name if shortened
+				n := itemResult.Get("name").String()
+				rev := buildReverseMapFromGeminiOriginal(originalRequestRawJSON)
+				if orig, ok := rev[n]; ok {
+					n = orig
+				}
+				functionCall, _ = sjson.Set(functionCall, "functionCall.name", n)
+			}
 
 			// Parse and set arguments
 			argsStr := itemResult.Get("arguments").String()
@@ -143,7 +151,7 @@ func ConvertCodexResponseToGemini(_ context.Context, modelName string, rawJSON [
 //
 // Returns:
 //   - string: A Gemini-compatible JSON response containing all message content and metadata
-func ConvertCodexResponseToGeminiNonStream(_ context.Context, modelName string, rawJSON []byte, _ *any) string {
+func ConvertCodexResponseToGeminiNonStream(_ context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, _ *any) string {
 	scanner := bufio.NewScanner(bytes.NewReader(rawJSON))
 	buffer := make([]byte, 10240*1024)
 	scanner.Buffer(buffer, 10240*1024)
@@ -250,7 +258,14 @@ func ConvertCodexResponseToGeminiNonStream(_ context.Context, modelName string,
 						hasToolCall = true
 						functionCall := map[string]interface{}{
 							"functionCall": map[string]interface{}{
-								"name": value.Get("name").String(),
+								"name": func() string {
+									n := value.Get("name").String()
+									rev := buildReverseMapFromGeminiOriginal(originalRequestRawJSON)
+									if orig, ok := rev[n]; ok {
+										return orig
+									}
+									return n
+								}(),
 								"args": map[string]interface{}{},
 							},
 						}
@@ -292,6 +307,35 @@ func ConvertCodexResponseToGeminiNonStream(_ context.Context, modelName string,
 	return ""
 }
 
+// buildReverseMapFromGeminiOriginal builds a map[short]original from original Gemini request tools.
+func buildReverseMapFromGeminiOriginal(original []byte) map[string]string {
+	tools := gjson.GetBytes(original, "tools")
+	rev := map[string]string{}
+	if !tools.IsArray() {
+		return rev
+	}
+	var names []string
+	tarr := tools.Array()
+	for i := 0; i < len(tarr); i++ {
+		fns := tarr[i].Get("functionDeclarations")
+		if !fns.IsArray() {
+			continue
+		}
+		for _, fn := range fns.Array() {
+			if v := fn.Get("name"); v.Exists() {
+				names = append(names, v.String())
+			}
+		}
+	}
+	if len(names) > 0 {
+		m := buildShortNameMap(names)
+		for orig, short := range m {
+			rev[short] = orig
+		}
+	}
+	return rev
+}
+
 // mustMarshalJSON marshals a value to JSON, panicking on error.
 func mustMarshalJSON(v interface{}) string {
 	data, err := json.Marshal(v)