Compare commits

...

52 Commits

Author SHA1 Message Date
Luis Pater
327cc7039e **refactor(auth): use customizable HTTP client for Antigravity requests**
- Replaced `http.DefaultClient` with a configurable `http.Client` instance for Antigravity OAuth flow methods.
- Updated `exchangeAntigravityCode` and `fetchAntigravityUserInfo` to accept `httpClient` as a parameter.
- Added `util.SetProxy` usage to initialize the `httpClient` with proxy support.
2025-11-21 20:54:56 +08:00
Luis Pater
b4d15ace91 Merge pull request #296 from router-for-me/antigravity
Antigravity bugfix
2025-11-21 17:32:36 +08:00
hkfires
abc2465b29 fix(gemini-cli): ignore thoughtSignature and empty parts 2025-11-21 17:12:56 +08:00
hkfires
4ba5b43d82 feat(executor): share SSE usage filtering across streams 2025-11-21 16:51:05 +08:00
hkfires
27faf718a3 fix(auth): use fixed antigravity callback port 51121 2025-11-21 13:56:33 +08:00
Luis Pater
2d84d2fb6a **feat(auth, executor, cmd): add Antigravity provider integration**
- Implemented OAuth login flow for the Antigravity provider in `auth/antigravity.go`.
- Added `AntigravityExecutor` for handling requests and streaming via Antigravity APIs.
- Created `antigravity_login.go` command for triggering Antigravity authentication.
- Introduced OpenAI-to-Antigravity translation logic in `translator/antigravity/openai/chat-completions`.

**refactor(translator, executor): update Gemini CLI response translation and add Antigravity payload customization**

- Renamed Gemini CLI translation methods to align with response handling (`ConvertGeminiCliResponseToGemini` and `ConvertGeminiCliResponseToGeminiNonStream`).
- Updated `init.go` to reflect these method changes.
- Introduced `geminiToAntigravity` function to embed metadata (`model`, `userAgent`, `project`, etc.) into Antigravity payloads.
- Added random project, request, and session ID generators for enhanced tracking.
- Streamlined `buildRequest` to use `geminiToAntigravity` transformation before request execution.
2025-11-21 12:43:16 +08:00
Luis Pater
cbcfeb92cc Fixed: #291
**feat(executor): add thinking level to budget conversion utility**

- Introduced `ConvertThinkingLevelToBudget` to map thinking level ("high"/"low") to corresponding budget values.
- Applied the utility in `aistudio_executor.go` before stripping unsupported configs.
- Updated dependencies to include `tidwall/gjson` for JSON parsing.
2025-11-21 00:48:12 +08:00
Luis Pater
db81331ae8 **refactor(middleware): extract request logging logic and optimize condition checks**
- Added `shouldLogRequest` helper to simplify path-based request logging logic.
- Updated middleware to skip management endpoints for improved security.
- Introduced an explicit `nil` logger check for minimal overhead.
- Updated dependencies in `go.mod`.

**feat(auth): add handling for 404 response with retry logic**

- Introduced support for 404 `not_found` status with a 12-hour backoff period.
- Updated `manager.go` to align state and status messages for 404 scenarios.

**refactor(translator): comment out debug logging in Gemini responses request**
2025-11-20 23:20:40 +08:00
Luis Pater
93fa1d1802 **docs: add Amp CLI integration guide to Chinese documentation**
- Updated `README_CN.md` to introduce Amp CLI and IDE support.
- Added detailed integration guide in `docs/amp-cli-integration_CN.md`.
- Covered setup, configuration, OAuth, security, and usage of Amp CLI with Google/ChatGPT/Claude subscriptions.
2025-11-20 21:07:20 +08:00
Luis Pater
b70bfd8092 Merge pull request #287 from ben-vargas/feat-amp-cli-module
Amp CLI Integration Module
2025-11-20 20:28:03 +08:00
Luis Pater
9ff38dd785 Merge branch 'dev' into feat-amp-cli-module 2025-11-20 20:26:47 +08:00
Luis Pater
98596c0a3f **refactor(translator): remove service_tier from Codex OpenAI request payload** 2025-11-20 20:12:06 +08:00
Luis Pater
670ce2e528 Merge pull request #285 from router-for-me/iflow
feat(iflow): add cookie-based authentication endpoint
2025-11-20 20:04:38 +08:00
hkfires
3f4f8b3b2d feat(iflow): add cookie-based authentication endpoint 2025-11-20 18:23:43 +08:00
Luis Pater
371324c090 **feat(registry): expand Gemini model definitions and support Vertex AI** 2025-11-20 18:16:26 +08:00
Luis Pater
d50b0f7524 **refactor(executor): simplify Gemini CLI execution and remove internal retry logic**
- Removed nested retry handling for 429 rate limit errors.
- Simplified request/response handling by cleaning redundant retry-related code.
- Eliminated `parseRetryDelay` function and max retry configuration logic.
2025-11-20 17:49:37 +08:00
Ben Vargas
a6cb16bb48 security: fix localhost middleware header spoofing vulnerability
Fix critical security vulnerability in amp-restrict-management-to-localhost
feature where attackers could bypass localhost restriction by spoofing
X-Forwarded-For headers.

Changes:
- Use RemoteAddr (actual TCP connection) instead of ClientIP() in
  localhostOnlyMiddleware to prevent header spoofing attacks
- Add comprehensive test coverage for spoofing prevention (6 test cases)
- Update documentation with reverse proxy deployment guidance and
  limitations of the RemoteAddr approach

The fix prevents attacks like:
  curl -H "X-Forwarded-For: 127.0.0.1" https://server/api/user

Trade-off: Users behind reverse proxies will need to disable the feature
and use alternative security measures (firewall rules, proxy ACLs).

Addresses security review feedback from PR #287.
2025-11-19 22:09:04 -07:00
Ben Vargas
70ee4e0aa0 chore: remove unused httpx sdk package 2025-11-19 21:17:52 -07:00
Ben Vargas
03334f8bb4 chore: revert gitignore change 2025-11-19 20:42:23 -07:00
Ben Vargas
5a2bebccfa fix: remove duplicate CountTokens stub 2025-11-19 20:00:39 -07:00
Luis Pater
0586da9c2b **refactor(registry): move Gemini 3 Pro Preview model definition to base set** 2025-11-20 10:51:16 +08:00
Ben Vargas
3d8d02bfc3 Fix amp v1beta1 routing and gemini retry config 2025-11-19 19:11:35 -07:00
Ben Vargas
7ae00320dc fix(amp): enable OAuth fallback for Gemini v1beta1 routes
AMP CLI sends Gemini requests to non-standard paths that were being
directly proxied to ampcode.com without checking for local OAuth.

This fix adds:
- GeminiBridge handler to transform AMP CLI paths to standard format
- Enhanced model extraction from AMP's /publishers/google/models/* paths
- FallbackHandler wrapper to check for local OAuth before proxying

Flow:
- If user has local Google OAuth → use it (free tier)
- If no local OAuth → fallback to ampcode.com (charges credits)

Fixes issue where gemini-3-pro-preview requests always charged AMP
credits even when user had valid Google Cloud OAuth configured.
2025-11-19 18:23:17 -07:00
Ben Vargas
1fb96f5379 docs: reposition Amp CLI as integrated feature for upstream PR
- Update README.md to present Amp CLI as standard feature, not fork differentiator
- Remove USING_WITH_FACTORY_AND_AMP.md (fork-specific, Factory docs live upstream)
- Add comprehensive docs/amp-cli-integration.md with setup, config, troubleshooting
- Eliminate fork justification messaging throughout documentation
- Prepare Amp CLI integration for upstream merge consideration

This positions Amp CLI support as a natural extension of CLIProxyAPI's
multi-client architecture rather than a fork-specific feature.
2025-11-19 18:23:17 -07:00
Ben Vargas
897d108e4c docs: update Factory config with GPT-5.1 models and explicit reasoning levels
- Replace deprecated GPT-5 and GPT-5-Codex with GPT-5.1 family
- Add explicit reasoning effort levels (low/medium/high)
- Remove duplicate base models (use medium as default)
- GPT-5.1 Codex Mini supports medium/high only (per OpenAI docs)
- Remove older Claude Sonnet 4 (keep 4.5)
- Final config: 11 models (3 Claude + 8 GPT-5.1 variants)
2025-11-19 18:23:17 -07:00
Ben Vargas
72d82268e5 fix(amp): filter context-1m beta header for local OAuth providers
Amp CLI sends 'context-1m-2025-08-07' in Anthropic-Beta header which
requires a special 1M context window subscription. After upstream rebase
to v6.3.7 (commit 38cfbac), CLIProxyAPI now respects client-provided
Anthropic-Beta headers instead of always using defaults.

When users configure local OAuth providers (Claude, etc), requests bypass
the ampcode.com proxy and use their own API subscriptions. These personal
subscriptions typically don't include the 1M context beta feature, causing
'long context beta not available' errors.

Changes:
- Add filterBetaFeatures() helper to strip specific beta features
- Filter context-1m-2025-08-07 in fallback handler when using local providers
- Preserve full headers when proxying to ampcode.com (paid users get all features)
- Add 7 test cases covering all edge cases

This fix is isolated to the Amp module and only affects the local provider
path. Users proxying through ampcode.com are unaffected and receive full
1M context support as part of their paid service.
2025-11-19 18:23:17 -07:00
Ben Vargas
8193392bfe Add AMP fallback proxy and shared Gemini normalization
- add fallback handler that forwards Amp provider requests to ampcode.com when the provider isn’t configured locally
- wrap AMP provider routes with the fallback so requests always have a handler
- share Gemini thinking model normalization helper between core handlers and AMP fallback
2025-11-19 18:23:17 -07:00
Ben Vargas
9ad0f3f91e feat: Add Amp CLI integration with comprehensive documentation
Add full Amp CLI support to enable routing AI model requests through the proxy
while maintaining Amp-specific features like thread management, user info, and
telemetry. Includes complete documentation and pull bot configuration.

Features:
- Modular architecture with RouteModule interface for clean integration
- Reverse proxy for Amp management routes (thread/user/meta/ads/telemetry)
- Provider-specific route aliases (/api/provider/{provider}/*)
- Secret management with precedence: config > env > file
- 5-minute secret caching to reduce file I/O
- Automatic gzip decompression for responses
- Proper connection cleanup to prevent leaks
- Localhost-only restriction for management routes (configurable)
- CORS protection for management endpoints

Documentation:
- Complete setup guide (USING_WITH_FACTORY_AND_AMP.md)
- OAuth setup for OpenAI (ChatGPT Plus/Pro) and Anthropic (Claude Pro/Max)
- Factory CLI config examples with all model variants
- Amp CLI/IDE configuration examples
- tmux setup for remote server deployment
- Screenshots and diagrams

Configuration:
- Pull bot disabled for this repo (manual rebase workflow)
- Config fields: AmpUpstreamURL, AmpUpstreamAPIKey, AmpRestrictManagementToLocalhost
- Compatible with upstream DisableCooling and other features

Technical details:
- internal/api/modules/amp/: Complete Amp routing module
- sdk/api/httpx/: HTTP utilities for gzip/transport
- 94.6% test coverage with 34 comprehensive test cases
- Clean integration minimizes merge conflict risk

Security:
- Management routes restricted to localhost by default
- Configurable via amp-restrict-management-to-localhost
- Prevents drive-by browser attacks on user data

This provides a production-ready foundation for Amp CLI integration while
maintaining clean separation from upstream code for easy rebasing.

Amp-Thread-ID: https://ampcode.com/threads/T-9e2befc5-f969-41c6-890c-5b779d58cf18
2025-11-19 18:23:17 -07:00
Luis Pater
618511ff67 Merge pull request #280 from ben-vargas/feat-enable-gemini-3-cli
feat: enable Gemini 3 Pro Preview with OAuth support
2025-11-20 08:46:57 +08:00
Ben Vargas
0ff094b87f fix(executor): prevent streaming on failed response when no fallback
Fix critical bug where ExecuteStream would create a streaming channel
from a failed (non-2xx) response after exhausting all retries with no
fallback models available.

When retries were exhausted on the last model, the code would break from
the inner loop but fall through to streaming channel creation (line 401),
immediately returning at line 461. This made the error handling code at
lines 464-471 unreachable, causing clients to receive an empty/closed
stream instead of a proper error response.

Solution: Check if httpResp is non-2xx before creating the streaming
channel. If failed, continue the outer loop to reach error handling.

Identified by: codex-bot review
Ref: https://github.com/router-for-me/CLIProxyAPI/pull/280#pullrequestreview-3484560423
2025-11-19 13:14:40 -07:00
Ben Vargas
ed23472d94 fix(executor): prevent streaming from 429 response when fallback available
Fix critical bug where ExecuteStream would create a streaming channel
using a 429 error response instead of continuing to the next fallback
model after exhausting retries.

When 429 retries were exhausted and a fallback model was available,
the inner retry loop would break but immediately fall through to the
streaming channel creation, attempting to stream from the failed 429
response instead of trying the next model.

Solution: Add shouldContinueToNextModel flag to explicitly skip the
streaming logic and continue the outer model loop when appropriate.

Identified by: codex-bot review
Ref: https://github.com/router-for-me/CLIProxyAPI/pull/280#pullrequestreview-3484479106
2025-11-19 13:05:38 -07:00
Ben Vargas
ede4471b84 feat(translator): add default thinkingConfig for gemini-3-pro-preview
Match official Gemini CLI behavior by always sending default
thinkingConfig when client doesn't specify reasoning parameters.

- Set thinkingBudget=-1 (dynamic) for gemini-3-pro-preview
- Set include_thoughts=true to return thinking process
- Apply to both /v1/chat/completions and /v1/responses endpoints
- See: ai-gemini-cli/packages/core/src/config/defaultModelConfigs.ts
2025-11-19 12:47:39 -07:00
Ben Vargas
6a3de3a89c feat(executor): add intelligent retry logic for 429 rate limits
Implement Google RetryInfo.retryDelay support for handling 429 rate
limit errors. Retries same model up to 3 times using exact delays
from Google's API before trying fallback models.

- Add parseRetryDelay() to extract Google's retry guidance
- Implement inner retry loop in Execute() and ExecuteStream()
- Context-aware waiting with cancellation support
- Cap delays at 60s maximum for safety
2025-11-19 12:47:39 -07:00
Ben Vargas
782bba0bc4 feat(registry): enable gemini-3-pro-preview for gemini-cli provider
Add gemini-3-pro-preview model to GetGeminiCLIModels() to make it
available for OAuth-based Gemini CLI users, matching the model
already available in AI Studio provider.

Model spec:
- ID: gemini-3-pro-preview
- Version: 3.0
- Input: 1M tokens
- Output: 64K tokens
- Thinking: 128-32K tokens (dynamic)
2025-11-19 12:47:39 -07:00
Luis Pater
bf116b68f8 **feat(registry): add GPT-5.1 Codex Max model definitions and support**
- Introduced `gpt-5.1-codex-max` variants to model definitions (`low`, `medium`, `high`, `xhigh`).
- Updated executor logic to map effort levels for Codex Max models.
- Added `lastCodexMaxPrompt` processing for `gpt-5.1-codex-max` prompts.
- Defined instructions for `gpt-5.1-codex-max` in a new file: `codex_instructions/gpt-5.1-codex-max_prompt.md`.
2025-11-20 03:12:22 +08:00
Luis Pater
cc3cf09c00 **feat(auth): add AuthIndex for diagnostics and ensure usage recording** 2025-11-19 22:02:40 +08:00
Luis Pater
9acfbcc2a0 Merge pull request #275 from router-for-me/iflow
Iflow
2025-11-19 20:44:54 +08:00
hkfires
b285b07986 fix(iflow): adjust auth filename email sanitization 2025-11-19 19:50:06 +08:00
Luis Pater
c40e00526b Merge pull request #274 from router-for-me/log
fix: detect HTML error bodies without text/html content type
2025-11-19 17:40:06 +08:00
hkfires
8a33f3ef69 fix: detect HTML error bodies without text/html content type 2025-11-19 14:45:33 +08:00
Luis Pater
7a8e00fcea **fix(translator): handle missing parameters in Gemini tool schema gracefully** 2025-11-19 13:19:46 +08:00
Luis Pater
89771216a1 **feat(translator): add ThoughtSignature handling in Gemini request transformations** 2025-11-19 11:34:13 +08:00
Luis Pater
14ddfd4b79 Merge pull request #270 from router-for-me/iflow
feat(auth): add iFlow cookie-based authentication support
2025-11-19 01:54:34 +08:00
Luis Pater
567227f35f Merge pull request #268 from router-for-me/tools
fix: use underscore suffix in short name mapping
2025-11-19 01:43:41 +08:00
Luis Pater
17016ae6a5 **feat(registry): add Gemini 3 Pro Preview model definition** 2025-11-18 23:48:21 +08:00
Luis Pater
01b7b60901 **feat(registry): add Gemini 3 Pro Preview model definition** 2025-11-18 23:46:58 +08:00
hkfires
b52a5cc066 feat(auth): add iFlow cookie-based authentication support 2025-11-18 22:35:35 +08:00
hkfires
1ba057112a fix: use underscore suffix in short name mapping
Replace the "~<n>" suffix with "_<n>" when generating unique short names in codex translators (Claude, Gemini, OpenAI chat).
This avoids using a special character in identifiers, improving compatibility with downstream APIs while preserving length constraints.
2025-11-18 16:59:25 +08:00
Luis Pater
23a7633e6d **fix(registry): update Thinking parameters and replace Gemini-3 Preview with Gemini-2.5 Flash Lite** 2025-11-18 11:51:52 +08:00
Luis Pater
e5e985978d Fixed: #263
**fix(translator): remove input_examples from tool schema in Gemini-Claude requests**
2025-11-18 11:27:48 +08:00
Luis Pater
db2d22c978 **fix(runtime): simplify scanner buffer allocation in executor implementations** 2025-11-18 10:59:49 +08:00
Luis Pater
1c815c58a6 **fix(translator): simplify string handling in Gemini responses** 2025-11-16 19:02:27 +08:00
70 changed files with 5835 additions and 327 deletions

View File

@@ -25,6 +25,7 @@ Get 10% OFF GLM CODING PLANhttps://z.ai/subscribe?ic=8JVLJQFSKB
- Claude Code support via OAuth login
- Qwen Code support via OAuth login
- iFlow support via OAuth login
- Amp CLI and IDE extensions support with provider routing
- Streaming and non-streaming responses
- Function calling/tools support
- Multimodal input support (text and images)
@@ -48,6 +49,17 @@ CLIProxyAPI Guides: [https://help.router-for.me/](https://help.router-for.me/)
see [MANAGEMENT_API.md](https://help.router-for.me/management/api)
## Amp CLI Support
CLIProxyAPI includes integrated support for [Amp CLI](https://ampcode.com) and Amp IDE extensions, enabling you to use your Google/ChatGPT/Claude OAuth subscriptions with Amp's coding tools:
- Provider route aliases for Amp's API patterns (`/api/provider/{provider}/v1...`)
- Management proxy for OAuth authentication and account features
- Smart model fallback with automatic routing
- Security-first design with localhost-only management endpoints
**→ [Complete Amp CLI Integration Guide](docs/amp-cli-integration.md)**
## SDK Docs
- Usage: [docs/sdk-usage.md](docs/sdk-usage.md)

View File

@@ -48,6 +48,17 @@ CLIProxyAPI 用户手册: [https://help.router-for.me/](https://help.router-fo
请参见 [MANAGEMENT_API_CN.md](https://help.router-for.me/cn/management/api)
## Amp CLI 支持
CLIProxyAPI 已内置对 [Amp CLI](https://ampcode.com) 和 Amp IDE 扩展的支持,可让你使用自己的 Google/ChatGPT/Claude OAuth 订阅来配合 Amp 编码工具:
- 提供商路由别名,兼容 Amp 的 API 路径模式(`/api/provider/{provider}/v1...`
- 管理代理,处理 OAuth 认证和账号功能
- 智能模型回退与自动路由
- 以安全为先的设计,管理端点仅限 localhost
**→ [Amp CLI 完整集成指南](docs/amp-cli-integration_CN.md)**
## SDK 文档
- 使用文档:[docs/sdk-usage_CN.md](docs/sdk-usage_CN.md)

View File

@@ -59,7 +59,9 @@ func main() {
var claudeLogin bool
var qwenLogin bool
var iflowLogin bool
var iflowCookie bool
var noBrowser bool
var antigravityLogin bool
var projectID string
var vertexImport string
var configPath string
@@ -71,7 +73,9 @@ func main() {
flag.BoolVar(&claudeLogin, "claude-login", false, "Login to Claude using OAuth")
flag.BoolVar(&qwenLogin, "qwen-login", false, "Login to Qwen using OAuth")
flag.BoolVar(&iflowLogin, "iflow-login", false, "Login to iFlow using OAuth")
flag.BoolVar(&iflowCookie, "iflow-cookie", false, "Login to iFlow using Cookie")
flag.BoolVar(&noBrowser, "no-browser", false, "Don't open browser automatically for OAuth")
flag.BoolVar(&antigravityLogin, "antigravity-login", false, "Login to Antigravity using OAuth")
flag.StringVar(&projectID, "project_id", "", "Project ID (Gemini only, not required)")
flag.StringVar(&configPath, "config", DefaultConfigPath, "Configure File Path")
flag.StringVar(&vertexImport, "vertex-import", "", "Import Vertex service account key JSON file")
@@ -429,6 +433,9 @@ func main() {
} else if login {
// Handle Google/Gemini login
cmd.DoLogin(cfg, projectID, options)
} else if antigravityLogin {
// Handle Antigravity login
cmd.DoAntigravityLogin(cfg, options)
} else if codexLogin {
// Handle Codex login
cmd.DoCodexLogin(cfg, options)
@@ -439,6 +446,8 @@ func main() {
cmd.DoQwenLogin(cfg, options)
} else if iflowLogin {
cmd.DoIFlowLogin(cfg, options)
} else if iflowCookie {
cmd.DoIFlowCookieAuth(cfg, options)
} else {
// In cloud deploy mode without config file, just wait for shutdown signals
if isCloudDeploy && !configFileExists {

View File

@@ -19,6 +19,7 @@ services:
- "8085:8085"
- "1455:1455"
- "54545:54545"
- "51121:51121"
- "11451:11451"
volumes:
- ./config.yaml:/CLIProxyAPI/config.yaml

392
docs/amp-cli-integration.md Normal file
View File

@@ -0,0 +1,392 @@
# Amp CLI Integration Guide
This guide explains how to use CLIProxyAPI with Amp CLI and Amp IDE extensions, enabling you to use your existing Google/ChatGPT/Claude subscriptions (via OAuth) with Amp's CLI.
## Table of Contents
- [Overview](#overview)
- [Which Providers Should You Authenticate?](#which-providers-should-you-authenticate)
- [Architecture](#architecture)
- [Configuration](#configuration)
- [Setup](#setup)
- [Usage](#usage)
- [Troubleshooting](#troubleshooting)
## Overview
The Amp CLI integration adds specialized routing to support Amp's API patterns while maintaining full compatibility with all existing CLIProxyAPI features. This allows you to use both traditional CLIProxyAPI features and Amp CLI with the same proxy server.
### Key Features
- **Provider route aliases**: Maps Amp's `/api/provider/{provider}/v1...` patterns to CLIProxyAPI handlers
- **Management proxy**: Forwards OAuth and account management requests to Amp's control plane
- **Smart fallback**: Automatically routes unconfigured models to ampcode.com
- **Secret management**: Configurable precedence (config > env > file) with 5-minute caching
- **Security-first**: Management routes restricted to localhost by default
- **Automatic gzip handling**: Decompresses responses from Amp upstream
### What You Can Do
- Use Amp CLI with your Google account (Gemini 3 Pro Preview, Gemini 2.5 Pro, Gemini 2.5 Flash)
- Use Amp CLI with your ChatGPT Plus/Pro subscription (GPT-5, GPT-5 Codex models)
- Use Amp CLI with your Claude Pro/Max subscription (Claude Sonnet 4.5, Opus 4.1)
- Use Amp IDE extensions (VS Code, Cursor, Windsurf, etc.) with the same proxy
- Run multiple CLI tools (Factory + Amp) through one proxy server
- Route unconfigured models automatically through ampcode.com
### Which Providers Should You Authenticate?
**Important**: The providers you need to authenticate depend on which models and features your installed version of Amp currently uses. Amp employs different providers for various agent modes and specialized subagents:
- **Smart mode**: Uses Google/Gemini models (Gemini 3 Pro)
- **Rush mode**: Uses Anthropic/Claude models (Claude Haiku 4.5)
- **Oracle subagent**: Uses OpenAI/GPT models (GPT-5 medium reasoning)
- **Librarian subagent**: Uses Anthropic/Claude models (Claude Sonnet 4.5)
- **Search subagent**: Uses Anthropic/Claude models (Claude Haiku 4.5)
- **Review feature**: Uses Google/Gemini models (Gemini 2.5 Flash-Lite)
For the most current information about which models Amp uses, see the **[Amp Models Documentation](https://ampcode.com/models)**.
#### Fallback Behavior
CLIProxyAPI uses a smart fallback system:
1. **Provider authenticated locally** (`--login`, `--codex-login`, `--claude-login`):
- Requests use **your OAuth subscription** (ChatGPT Plus/Pro, Claude Pro/Max, Google account)
- You benefit from your subscription's included usage quotas
- No Amp credits consumed
2. **Provider NOT authenticated locally**:
- Requests automatically forward to **ampcode.com**
- Uses Amp's backend provider connections
- **Requires Amp credits** if the provider is paid (OpenAI, Anthropic paid tiers)
- May result in errors if Amp credit balance is insufficient
**Recommendation**: Authenticate all providers you have subscriptions for to maximize value and minimize Amp credit usage. If you don't have subscriptions to all providers Amp uses, ensure you have sufficient Amp credits available for fallback requests.
## Architecture
### Request Flow
```
Amp CLI/IDE
├─ Provider API requests (/api/provider/{provider}/v1/...)
│ ↓
│ ├─ Model configured locally?
│ │ YES → Use local OAuth tokens (OpenAI/Claude/Gemini handlers)
│ │ NO → Forward to ampcode.com (reverse proxy)
│ ↓
│ Response
└─ Management requests (/api/auth, /api/user, /api/threads, ...)
├─ Localhost check (security)
└─ Reverse proxy to ampcode.com
Response (auto-decompressed if gzipped)
```
### Components
The Amp integration is implemented as a modular routing module (`internal/api/modules/amp/`) with these components:
1. **Route Aliases** (`routes.go`): Maps Amp-style paths to standard handlers
2. **Reverse Proxy** (`proxy.go`): Forwards management requests to ampcode.com
3. **Fallback Handler** (`fallback_handlers.go`): Routes unconfigured models to ampcode.com
4. **Secret Management** (`secret.go`): Multi-source API key resolution with caching
5. **Main Module** (`amp.go`): Orchestrates registration and configuration
## Configuration
### Basic Configuration
Add these fields to your `config.yaml`:
```yaml
# Amp upstream control plane (required for management routes)
amp-upstream-url: "https://ampcode.com"
# Optional: Override API key (otherwise uses env or file)
# amp-upstream-api-key: "your-amp-api-key"
# Security: restrict management routes to localhost (recommended)
amp-restrict-management-to-localhost: true
```
### Secret Resolution Precedence
The Amp module resolves API keys using this precedence order:
| Source | Key | Priority | Cache |
|--------|-----|----------|-------|
| Config file | `amp-upstream-api-key` | High | No |
| Environment | `AMP_API_KEY` | Medium | No |
| Amp secrets file | `~/.local/share/amp/secrets.json` | Low | 5 min |
**Recommendation**: Use the Amp secrets file (lowest precedence) for normal usage. This file is automatically managed by `amp login`.
### Security Settings
**`amp-restrict-management-to-localhost`** (default: `true`)
When enabled, management routes (`/api/auth`, `/api/user`, `/api/threads`, etc.) only accept connections from localhost (127.0.0.1, ::1). This prevents:
- Drive-by browser attacks
- Remote access to management endpoints
- CORS-based attacks
- Header spoofing attacks (e.g., `X-Forwarded-For: 127.0.0.1`)
#### How It Works
This restriction uses the **actual TCP connection address** (`RemoteAddr`), not HTTP headers like `X-Forwarded-For`. This prevents header spoofing attacks but has important implications:
-**Works for direct connections**: Running CLIProxyAPI directly on your machine or server
- ⚠️ **May not work behind reverse proxies**: If deploying behind nginx, Cloudflare, or other proxies, the connection will appear to come from the proxy's IP, not localhost
#### Reverse Proxy Deployments
If you need to run CLIProxyAPI behind a reverse proxy (nginx, Caddy, Cloudflare Tunnel, etc.):
1. **Disable the localhost restriction**:
```yaml
amp-restrict-management-to-localhost: false
```
2. **Use alternative security measures**:
- Firewall rules restricting access to management routes
- Proxy-level authentication (HTTP Basic Auth, OAuth)
- Network-level isolation (VPN, Tailscale, Cloudflare Access)
- Bind CLIProxyAPI to `127.0.0.1` only and access via SSH tunnel
3. **Example nginx configuration** (blocks external access to management routes):
```nginx
location /api/auth { deny all; }
location /api/user { deny all; }
location /api/threads { deny all; }
location /api/internal { deny all; }
```
**Important**: Only disable `amp-restrict-management-to-localhost` if you understand the security implications and have other protections in place.
## Setup
### 1. Configure CLIProxyAPI
Create or edit `config.yaml`:
```yaml
port: 8317
auth-dir: "~/.cli-proxy-api"
# Amp integration
amp-upstream-url: "https://ampcode.com"
amp-restrict-management-to-localhost: true
# Other standard settings...
debug: false
logging-to-file: true
```
### 2. Authenticate with Providers
Run OAuth login for the providers you want to use:
**Google Account (Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 3 Pro Preview):**
```bash
./cli-proxy-api --login
```
**ChatGPT Plus/Pro (GPT-5, GPT-5 Codex):**
```bash
./cli-proxy-api --codex-login
```
**Claude Pro/Max (Claude Sonnet 4.5, Opus 4.1):**
```bash
./cli-proxy-api --claude-login
```
Tokens are saved to:
- Gemini: `~/.cli-proxy-api/gemini-<email>.json`
- OpenAI Codex: `~/.cli-proxy-api/codex-<email>.json`
- Claude: `~/.cli-proxy-api/claude-<email>.json`
### 3. Start the Proxy
```bash
./cli-proxy-api --config config.yaml
```
Or run in background with tmux (recommended for remote servers):
```bash
tmux new-session -d -s proxy "./cli-proxy-api --config config.yaml"
```
### 4. Configure Amp CLI
#### Option A: Settings File
Edit `~/.config/amp/settings.json`:
```json
{
"amp.url": "http://localhost:8317"
}
```
#### Option B: Environment Variable
```bash
export AMP_URL=http://localhost:8317
```
### 5. Login and Use Amp
Login through the proxy (proxied to ampcode.com):
```bash
amp login
```
Use Amp as normal:
```bash
amp "Write a hello world program in Python"
```
### 6. (Optional) Configure Amp IDE Extension
The proxy also works with Amp IDE extensions for VS Code, Cursor, Windsurf, etc.
1. Open Amp extension settings in your IDE
2. Set **Amp URL** to `http://localhost:8317`
3. Login with your Amp account
4. Start using Amp in your IDE
Both CLI and IDE can use the proxy simultaneously.
## Usage
### Supported Routes
#### Provider Aliases (Always Available)
These routes work even without `amp-upstream-url` configured:
- `/api/provider/openai/v1/chat/completions`
- `/api/provider/openai/v1/responses`
- `/api/provider/anthropic/v1/messages`
- `/api/provider/google/v1beta/models/:action`
Amp CLI calls these routes with your OAuth-authenticated models configured in CLIProxyAPI.
#### Management Routes (Require `amp-upstream-url`)
These routes are proxied to ampcode.com:
- `/api/auth` - Authentication
- `/api/user` - User profile
- `/api/meta` - Metadata
- `/api/threads` - Conversation threads
- `/api/telemetry` - Usage telemetry
- `/api/internal` - Internal APIs
**Security**: Restricted to localhost by default.
### Model Fallback Behavior
When Amp requests a model:
1. **Check local configuration**: Does CLIProxyAPI have OAuth tokens for this model's provider?
2. **If YES**: Route to local handler (use your OAuth subscription)
3. **If NO**: Forward to ampcode.com (use Amp's default routing)
This enables seamless mixed usage:
- Models you've configured (Gemini, ChatGPT, Claude) → Your OAuth subscriptions
- Models you haven't configured → Amp's default providers
### Example API Calls
**Chat completion with local OAuth:**
```bash
curl http://localhost:8317/api/provider/openai/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-5",
"messages": [{"role": "user", "content": "Hello"}]
}'
```
**Management endpoint (localhost only):**
```bash
curl http://localhost:8317/api/user
```
## Troubleshooting
### Common Issues
| Symptom | Likely Cause | Fix |
|---------|--------------|-----|
| 404 on `/api/provider/...` | Incorrect route path | Ensure exact path: `/api/provider/{provider}/v1...` |
| 403 on `/api/user` | Non-localhost request | Run from same machine or disable `amp-restrict-management-to-localhost` (not recommended) |
| 401/403 from provider | Missing/expired OAuth | Re-run `--codex-login` or `--claude-login` |
| Amp gzip errors | Response decompression issue | Update to latest build; auto-decompression should handle this |
| Models not using proxy | Wrong Amp URL | Verify `amp.url` setting or `AMP_URL` environment variable |
| CORS errors | Protected management endpoint | Use CLI/terminal, not browser |
### Diagnostics
**Check proxy logs:**
```bash
# If logging-to-file: true
tail -f logs/requests.log
# If running in tmux
tmux attach-session -t proxy
```
**Enable debug mode** (temporarily):
```yaml
debug: true
```
**Test basic connectivity:**
```bash
# Check if proxy is running
curl http://localhost:8317/v1/models
# Check Amp-specific route
curl http://localhost:8317/api/provider/openai/v1/models
```
**Verify Amp configuration:**
```bash
# Check if Amp is using proxy
amp config get amp.url
# Or check environment
echo $AMP_URL
```
### Security Checklist
- ✅ Keep `amp-restrict-management-to-localhost: true` (default)
- ✅ Don't expose proxy publicly (bind to localhost or use firewall/VPN)
- ✅ Use the Amp secrets file (`~/.local/share/amp/secrets.json`) managed by `amp login`
- ✅ Rotate OAuth tokens periodically by re-running login commands
- ✅ Store config and auth-dir on encrypted disk if handling sensitive data
- ✅ Keep proxy binary up to date for security fixes
## Additional Resources
- [CLIProxyAPI Main Documentation](https://help.router-for.me/)
- [Amp CLI Official Manual](https://ampcode.com/manual)
- [Management API Reference](https://help.router-for.me/management/api)
- [SDK Documentation](sdk-usage.md)
## Disclaimer
This integration is for personal/educational use. Using reverse proxies or alternate API bases may violate provider Terms of Service. You are solely responsible for how you use this software. Accounts may be rate-limited, locked, or banned. No warranties. Use at your own risk.

View File

@@ -0,0 +1,392 @@
# Amp CLI 集成指南
本指南说明如何在 Amp CLI 和 Amp IDE 扩展中使用 CLIProxyAPI通过 OAuth 让你能够把已有的 Google/ChatGPT/Claude 订阅与 Amp 的 CLI 一起使用。
## 目录
- [概述](#概述)
- [应该认证哪些服务提供商?](#应该认证哪些服务提供商)
- [架构](#架构)
- [配置](#配置)
- [设置](#设置)
- [用法](#用法)
- [故障排查](#故障排查)
## 概述
Amp CLI 集成为 Amp 的 API 模式添加了专用路由,同时保持与现有 CLIProxyAPI 功能的完全兼容。这样你可以在同一个代理服务器上同时使用传统 CLIProxyAPI 功能和 Amp CLI。
### 主要特性
- **提供者路由别名**:将 Amp 的 `/api/provider/{provider}/v1...` 路径映射到 CLIProxyAPI 处理器
- **管理代理**:将 OAuth 和账号管理请求转发到 Amp 控制平面
- **智能回退**:自动将未配置的模型路由到 ampcode.com
- **密钥管理**:可配置优先级(配置 > 环境变量 > 文件),缓存 5 分钟
- **安全优先**:管理路由默认限制为 localhost
- **自动 gzip 处理**:自动解压来自 Amp 上游的响应
### 你可以做什么
- 使用 Amp CLI 搭配你的 Google 账号Gemini 3 Pro Preview、Gemini 2.5 Pro、Gemini 2.5 Flash
- 使用 Amp CLI 搭配你的 ChatGPT Plus/Pro 订阅GPT-5、GPT-5 Codex 模型)
- 使用 Amp CLI 搭配你的 Claude Pro/Max 订阅Claude Sonnet 4.5、Opus 4.1
- 将 Amp IDE 扩展VS Code、Cursor、Windsurf 等)与同一个代理一起使用
- 通过一个代理同时运行多个 CLI 工具Factory + Amp
- 将未配置的模型自动路由到 ampcode.com
### 应该认证哪些服务提供商?
**重要**:需要认证的提供商取决于你安装的 Amp 版本当前使用的模型和功能。Amp 的不同智能模式和子代理会使用不同的提供商:
- **Smart 模式**:使用 Google/Gemini 模型Gemini 3 Pro
- **Rush 模式**:使用 Anthropic/Claude 模型Claude Haiku 4.5
- **Oracle 子代理**:使用 OpenAI/GPT 模型GPT-5 medium reasoning
- **Librarian 子代理**:使用 Anthropic/Claude 模型Claude Sonnet 4.5
- **Search 子代理**:使用 Anthropic/Claude 模型Claude Haiku 4.5
- **Review 功能**:使用 Google/Gemini 模型Gemini 2.5 Flash-Lite
有关 Amp 当前使用哪些模型的最新信息,请参阅 **[Amp 模型文档](https://ampcode.com/models)**。
#### 回退行为
CLIProxyAPI 采用智能回退机制:
1. **本地已认证提供商**`--login``--codex-login``--claude-login`
- 请求使用**你的 OAuth 订阅**ChatGPT Plus/Pro、Claude Pro/Max、Google 账号)
- 享受订阅自带的额度
- 不消耗 Amp 额度
2. **本地未认证提供商**
- 请求自动转发到 **ampcode.com**
- 使用 Amp 的后端提供商连接
- 如果提供商是付费的OpenAI、Anthropic 付费档),**需要消耗 Amp 额度**
- 若 Amp 额度不足,可能产生错误
**建议**:对你有订阅的所有提供商都进行认证,以最大化价值并尽量减少 Amp 额度消耗。如果没有覆盖 Amp 使用的全部提供商,请确保为回退请求准备足够的 Amp 额度。
## 架构
### 请求流
```
Amp CLI/IDE
├─ Provider API requests (/api/provider/{provider}/v1/...)
│ ↓
│ ├─ Model configured locally?
│ │ YES → Use local OAuth tokens (OpenAI/Claude/Gemini handlers)
│ │ NO → Forward to ampcode.com (reverse proxy)
│ ↓
│ Response
└─ Management requests (/api/auth, /api/user, /api/threads, ...)
├─ Localhost check (security)
└─ Reverse proxy to ampcode.com
Response (auto-decompressed if gzipped)
```
### 组件
Amp 集成以模块化路由模块(`internal/api/modules/amp/`)实现,包含以下组件:
1. **路由别名**`routes.go`):将 Amp 风格的路径映射到标准处理器
2. **反向代理**`proxy.go`):将管理请求转发到 ampcode.com
3. **回退处理器**`fallback_handlers.go`):将未配置的模型路由到 ampcode.com
4. **密钥管理**`secret.go`):多来源 API 密钥解析并带缓存
5. **主模块**`amp.go`):负责注册和配置
## 配置
### 基础配置
`config.yaml` 中新增以下字段:
```yaml
# Amp 上游控制平面(管理路由必需)
amp-upstream-url: "https://ampcode.com"
# 可选:覆盖 API key否则使用环境变量或文件
# amp-upstream-api-key: "your-amp-api-key"
# 安全性:将管理路由限制为 localhost推荐
amp-restrict-management-to-localhost: true
```
### 密钥解析优先级
Amp 模块以如下优先级解析 API key
| 来源 | 键名 | 优先级 | 缓存 |
|------|------|--------|------|
| 配置文件 | `amp-upstream-api-key` | 高 | 无 |
| 环境变量 | `AMP_API_KEY` | 中 | 无 |
| Amp 密钥文件 | `~/.local/share/amp/secrets.json` | 低 | 5 分钟 |
**建议**:日常使用时采用 Amp 密钥文件(最低优先级)。该文件由 `amp login` 自动管理。
### 安全设置
**`amp-restrict-management-to-localhost`**(默认:`true`
启用后,管理路由(`/api/auth``/api/user``/api/threads` 等)只接受来自 localhost127.0.0.1、::1的连接可防止
- 浏览器探测式攻击
- 对管理端点的远程访问
- 基于 CORS 的攻击
- 伪造头攻击(例如 `X-Forwarded-For: 127.0.0.1`
#### 工作原理
此限制使用**实际的 TCP 连接地址**`RemoteAddr`),而非 `X-Forwarded-For` 等 HTTP 头,能防止头部伪造,但有重要影响:
-**直接连接可用**:在本机或服务器直接运行 CLIProxyAPI 时适用
- ⚠️ **可能不适用于反向代理场景**:部署在 nginx、Cloudflare 等代理后,请求源会显示为代理 IP 而非 localhost
#### 反向代理部署
若需要在反向代理nginx、Caddy、Cloudflare Tunnel 等)后运行 CLIProxyAPI
1. **关闭 localhost 限制**
```yaml
amp-restrict-management-to-localhost: false
```
2. **使用替代安全措施**
- 防火墙规则限制管理路由访问
- 代理层认证HTTP Basic Auth、OAuth
- 网络隔离VPN、Tailscale、Cloudflare Access
- 将 CLIProxyAPI 仅绑定 `127.0.0.1`,并通过 SSH 隧道访问
3. **nginx 示例配置**(阻止外部访问管理路由):
```nginx
location /api/auth { deny all; }
location /api/user { deny all; }
location /api/threads { deny all; }
location /api/internal { deny all; }
```
**重要**:只有在理解安全影响并已采取其他防护措施时,才关闭 `amp-restrict-management-to-localhost`。
## 设置
### 1. 配置 CLIProxyAPI
创建或编辑 `config.yaml`
```yaml
port: 8317
auth-dir: "~/.cli-proxy-api"
# Amp 集成
amp-upstream-url: "https://ampcode.com"
amp-restrict-management-to-localhost: true
# 其他常规设置...
debug: false
logging-to-file: true
```
### 2. 认证提供商
为要使用的提供商执行 OAuth 登录:
**Google 账号Gemini 2.5 Pro、Gemini 2.5 Flash、Gemini 3 Pro Preview**
```bash
./cli-proxy-api --login
```
**ChatGPT Plus/ProGPT-5、GPT-5 Codex**
```bash
./cli-proxy-api --codex-login
```
**Claude Pro/MaxClaude Sonnet 4.5、Opus 4.1**
```bash
./cli-proxy-api --claude-login
```
令牌会保存到:
- Gemini: `~/.cli-proxy-api/gemini-<email>.json`
- OpenAI Codex: `~/.cli-proxy-api/codex-<email>.json`
- Claude: `~/.cli-proxy-api/claude-<email>.json`
### 3. 启动代理
```bash
./cli-proxy-api --config config.yaml
```
或使用 tmux 在后台运行(推荐用于远程服务器):
```bash
tmux new-session -d -s proxy "./cli-proxy-api --config config.yaml"
```
### 4. 配置 Amp CLI
#### 方案 A配置文件
编辑 `~/.config/amp/settings.json`
```json
{
"amp.url": "http://localhost:8317"
}
```
#### 方案 B环境变量
```bash
export AMP_URL=http://localhost:8317
```
### 5. 登录并使用 Amp
通过代理登录(请求会被代理到 ampcode.com
```bash
amp login
```
像平常一样使用 Amp
```bash
amp "Write a hello world program in Python"
```
### 6. (可选)配置 Amp IDE 扩展
该代理同样适用于 VS Code、Cursor、Windsurf 等 Amp IDE 扩展。
1. 在 IDE 中打开 Amp 扩展设置
2. 将 **Amp URL** 设置为 `http://localhost:8317`
3. 用你的 Amp 账号登录
4. 在 IDE 中开始使用 Amp
CLI 和 IDE 可同时使用该代理。
## 用法
### 支持的路由
#### 提供商别名(始终可用)
这些路由即使未配置 `amp-upstream-url` 也可使用:
- `/api/provider/openai/v1/chat/completions`
- `/api/provider/openai/v1/responses`
- `/api/provider/anthropic/v1/messages`
- `/api/provider/google/v1beta/models/:action`
Amp CLI 会使用你在 CLIProxyAPI 中通过 OAuth 认证的模型来调用这些路由。
#### 管理路由(需要 `amp-upstream-url`
这些路由会被代理到 ampcode.com
- `/api/auth` - 认证
- `/api/user` - 用户资料
- `/api/meta` - 元数据
- `/api/threads` - 会话线程
- `/api/telemetry` - 使用遥测
- `/api/internal` - 内部 API
**安全性**:默认限制为 localhost。
### 模型回退行为
当 Amp 请求模型时:
1. **检查本地配置**CLIProxyAPI 是否有该模型提供商的 OAuth 令牌?
2. **如果有**:路由到本地处理器(使用你的 OAuth 订阅)
3. **如果没有**:转发到 ampcode.com使用 Amp 的默认路由)
这实现了无缝混用:
- 你已配置的模型Gemini、ChatGPT、Claude→ 你的 OAuth 订阅
- 未配置的模型 → Amp 的默认提供商
### 示例 API 调用
**使用本地 OAuth 的聊天补全:**
```bash
curl http://localhost:8317/api/provider/openai/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-5",
"messages": [{"role": "user", "content": "Hello"}]
}'
```
**管理端点(仅限 localhost**
```bash
curl http://localhost:8317/api/user
```
## 故障排查
### 常见问题
| 症状 | 可能原因 | 解决方案 |
|------|----------|----------|
| `/api/provider/...` 返回 404 | 路径错误 | 确保路径准确:`/api/provider/{provider}/v1...` |
| `/api/user` 返回 403 | 非 localhost 请求 | 在同一机器上访问,或关闭 `amp-restrict-management-to-localhost`(不推荐) |
| 提供商返回 401/403 | OAuth 缺失或过期 | 重新运行 `--codex-login` 或 `--claude-login` |
| Amp gzip 错误 | 响应解压问题 | 更新到最新构建;自动解压应能处理 |
| 模型未走代理 | Amp URL 设置错误 | 检查 `amp.url` 设置或 `AMP_URL` 环境变量 |
| CORS 错误 | 受保护的管理端点 | 使用 CLI/终端而非浏览器 |
### 诊断
**查看代理日志:**
```bash
# 若 logging-to-file: true
tail -f logs/requests.log
# 若运行在 tmux 中
tmux attach-session -t proxy
```
**临时开启调试模式:**
```yaml
debug: true
```
**测试基础连通性:**
```bash
# 检查代理是否运行
curl http://localhost:8317/v1/models
# 检查 Amp 特定路由
curl http://localhost:8317/api/provider/openai/v1/models
```
**验证 Amp 配置:**
```bash
# 检查 Amp 是否使用代理
amp config get amp.url
# 或检查环境变量
echo $AMP_URL
```
### 安全清单
- ✅ 保持 `amp-restrict-management-to-localhost: true`(默认)
- ✅ 不要将代理暴露到公共网络(绑定到 localhost 或使用防火墙/VPN
- ✅ 使用 `amp login` 管理的 Amp 密钥文件(`~/.local/share/amp/secrets.json`
- ✅ 定期重新登录轮换 OAuth 令牌
- ✅ 若处理敏感数据,使用加密磁盘存储配置和 auth-dir
- ✅ 保持代理二进制为最新版本以获取安全修复
## 其他资源
- [CLIProxyAPI 主文档](https://help.router-for.me/)
- [Amp CLI 官方手册](https://ampcode.com/manual)
- [管理 API 参考](https://help.router-for.me/management/api)
- [SDK 文档](sdk-usage.md)
## 免责声明
此集成仅用于个人或教育用途。使用反向代理或替代 API 基址可能违反提供商的服务条款。你需要对自己的使用方式负责。账号可能会被限速、锁定或封禁。软件不附带任何保证,使用风险自负。

View File

@@ -137,6 +137,10 @@ func (MyExecutor) Execute(ctx context.Context, a *coreauth.Auth, req clipexec.Re
return clipexec.Response{Payload: body}, nil
}
func (MyExecutor) CountTokens(context.Context, *coreauth.Auth, clipexec.Request, clipexec.Options) (clipexec.Response, error) {
return clipexec.Response{}, errors.New("count tokens not implemented")
}
func (MyExecutor) ExecuteStream(ctx context.Context, a *coreauth.Auth, req clipexec.Request, opts clipexec.Options) (<-chan clipexec.StreamChunk, error) {
ch := make(chan clipexec.StreamChunk, 1)
go func() {
@@ -146,10 +150,6 @@ func (MyExecutor) ExecuteStream(ctx context.Context, a *coreauth.Auth, req clipe
return ch, nil
}
func (MyExecutor) CountTokens(ctx context.Context, a *coreauth.Auth, req clipexec.Request, opts clipexec.Options) (clipexec.Response, error) {
return clipexec.Response{}, errors.New("not implemented")
}
func (MyExecutor) Refresh(ctx context.Context, a *coreauth.Auth) (*coreauth.Auth, error) {
return a, nil
}

2
go.mod
View File

@@ -3,6 +3,7 @@ module github.com/router-for-me/CLIProxyAPI/v6
go 1.24.0
require (
github.com/andybalholm/brotli v1.0.6
github.com/fsnotify/fsnotify v1.9.0
github.com/gin-gonic/gin v1.10.1
github.com/go-git/go-git/v6 v6.0.0-20251009132922-75a182125145
@@ -28,7 +29,6 @@ require (
cloud.google.com/go/compute/metadata v0.3.0 // indirect
github.com/Microsoft/go-winio v0.6.2 // indirect
github.com/ProtonMail/go-crypto v1.3.0 // indirect
github.com/andybalholm/brotli v1.0.6 // indirect
github.com/bytedance/sonic v1.11.6 // indirect
github.com/bytedance/sonic/loader v0.1.1 // indirect
github.com/cloudflare/circl v1.6.1 // indirect

View File

@@ -292,6 +292,7 @@ func (h *Handler) buildAuthFileEntry(auth *coreauth.Auth) gin.H {
if auth == nil {
return nil
}
auth.EnsureIndex()
runtimeOnly := isRuntimeOnlyAuth(auth)
if runtimeOnly && (auth.Disabled || auth.Status == coreauth.StatusDisabled) {
return nil
@@ -306,6 +307,7 @@ func (h *Handler) buildAuthFileEntry(auth *coreauth.Auth) gin.H {
}
entry := gin.H{
"id": auth.ID,
"auth_index": auth.Index,
"name": name,
"type": strings.TrimSpace(auth.Provider),
"provider": strings.TrimSpace(auth.Provider),
@@ -1441,6 +1443,87 @@ func (h *Handler) RequestIFlowToken(c *gin.Context) {
c.JSON(http.StatusOK, gin.H{"status": "ok", "url": authURL, "state": state})
}
func (h *Handler) RequestIFlowCookieToken(c *gin.Context) {
ctx := context.Background()
var payload struct {
Cookie string `json:"cookie"`
}
if err := c.ShouldBindJSON(&payload); err != nil {
c.JSON(http.StatusBadRequest, gin.H{"status": "error", "error": "cookie is required"})
return
}
cookieValue := strings.TrimSpace(payload.Cookie)
if cookieValue == "" {
c.JSON(http.StatusBadRequest, gin.H{"status": "error", "error": "cookie is required"})
return
}
cookieValue, errNormalize := iflowauth.NormalizeCookie(cookieValue)
if errNormalize != nil {
c.JSON(http.StatusBadRequest, gin.H{"status": "error", "error": errNormalize.Error()})
return
}
authSvc := iflowauth.NewIFlowAuth(h.cfg)
tokenData, errAuth := authSvc.AuthenticateWithCookie(ctx, cookieValue)
if errAuth != nil {
c.JSON(http.StatusBadRequest, gin.H{"status": "error", "error": errAuth.Error()})
return
}
tokenData.Cookie = cookieValue
tokenStorage := authSvc.CreateCookieTokenStorage(tokenData)
email := strings.TrimSpace(tokenStorage.Email)
if email == "" {
c.JSON(http.StatusBadRequest, gin.H{"status": "error", "error": "failed to extract email from token"})
return
}
fileName := iflowauth.SanitizeIFlowFileName(email)
if fileName == "" {
fileName = fmt.Sprintf("iflow-%d", time.Now().UnixMilli())
}
tokenStorage.Email = email
record := &coreauth.Auth{
ID: fmt.Sprintf("iflow-%s.json", fileName),
Provider: "iflow",
FileName: fmt.Sprintf("iflow-%s.json", fileName),
Storage: tokenStorage,
Metadata: map[string]any{
"email": email,
"api_key": tokenStorage.APIKey,
"expired": tokenStorage.Expire,
"cookie": tokenStorage.Cookie,
"type": tokenStorage.Type,
"last_refresh": tokenStorage.LastRefresh,
},
Attributes: map[string]string{
"api_key": tokenStorage.APIKey,
},
}
savedPath, errSave := h.saveTokenRecord(ctx, record)
if errSave != nil {
c.JSON(http.StatusInternalServerError, gin.H{"status": "error", "error": "failed to save authentication tokens"})
return
}
fmt.Printf("iFlow cookie authentication successful. Token saved to %s\n", savedPath)
c.JSON(http.StatusOK, gin.H{
"status": "ok",
"saved_path": savedPath,
"email": email,
"expired": tokenStorage.Expire,
"type": tokenStorage.Type,
})
}
type projectSelectionRequiredError struct{}
func (e *projectSelectionRequiredError) Error() string {

View File

@@ -19,13 +19,13 @@ import (
// logger, the middleware has minimal overhead.
func RequestLoggingMiddleware(logger logging.RequestLogger) gin.HandlerFunc {
return func(c *gin.Context) {
path := c.Request.URL.Path
shouldLog := false
if strings.HasPrefix(path, "/v1") {
shouldLog = true
if logger == nil {
c.Next()
return
}
if !shouldLog {
path := c.Request.URL.Path
if !shouldLogRequest(path) {
c.Next()
return
}
@@ -101,3 +101,13 @@ func captureRequestInfo(c *gin.Context) (*RequestInfo, error) {
Body: body,
}, nil
}
// shouldLogRequest determines whether the request should be logged.
// It skips management endpoints to avoid leaking secrets but allows
// all other routes, including module-provided ones, to honor request-log.
func shouldLogRequest(path string) bool {
if strings.HasPrefix(path, "/v0/management") || strings.HasPrefix(path, "/management") {
return false
}
return true
}

View File

@@ -0,0 +1,185 @@
// Package amp implements the Amp CLI routing module, providing OAuth-based
// integration with Amp CLI for ChatGPT and Anthropic subscriptions.
package amp
import (
"fmt"
"net/http/httputil"
"strings"
"sync"
"github.com/gin-gonic/gin"
"github.com/router-for-me/CLIProxyAPI/v6/internal/api/modules"
"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
sdkaccess "github.com/router-for-me/CLIProxyAPI/v6/sdk/access"
log "github.com/sirupsen/logrus"
)
// Option configures the AmpModule.
type Option func(*AmpModule)
// AmpModule implements the RouteModuleV2 interface for Amp CLI integration.
// It provides:
// - Reverse proxy to Amp control plane for OAuth/management
// - Provider-specific route aliases (/api/provider/{provider}/...)
// - Automatic gzip decompression for misconfigured upstreams
type AmpModule struct {
secretSource SecretSource
proxy *httputil.ReverseProxy
accessManager *sdkaccess.Manager
authMiddleware_ gin.HandlerFunc
enabled bool
registerOnce sync.Once
}
// New creates a new Amp routing module with the given options.
// This is the preferred constructor using the Option pattern.
//
// Example:
//
// ampModule := amp.New(
// amp.WithAccessManager(accessManager),
// amp.WithAuthMiddleware(authMiddleware),
// amp.WithSecretSource(customSecret),
// )
func New(opts ...Option) *AmpModule {
m := &AmpModule{
secretSource: nil, // Will be created on demand if not provided
}
for _, opt := range opts {
opt(m)
}
return m
}
// NewLegacy creates a new Amp routing module using the legacy constructor signature.
// This is provided for backwards compatibility.
//
// DEPRECATED: Use New with options instead.
func NewLegacy(accessManager *sdkaccess.Manager, authMiddleware gin.HandlerFunc) *AmpModule {
return New(
WithAccessManager(accessManager),
WithAuthMiddleware(authMiddleware),
)
}
// WithSecretSource sets a custom secret source for the module.
func WithSecretSource(source SecretSource) Option {
return func(m *AmpModule) {
m.secretSource = source
}
}
// WithAccessManager sets the access manager for the module.
func WithAccessManager(am *sdkaccess.Manager) Option {
return func(m *AmpModule) {
m.accessManager = am
}
}
// WithAuthMiddleware sets the authentication middleware for provider routes.
func WithAuthMiddleware(middleware gin.HandlerFunc) Option {
return func(m *AmpModule) {
m.authMiddleware_ = middleware
}
}
// Name returns the module identifier
func (m *AmpModule) Name() string {
return "amp-routing"
}
// Register sets up Amp routes if configured.
// This implements the RouteModuleV2 interface with Context.
// Routes are registered only once via sync.Once for idempotent behavior.
func (m *AmpModule) Register(ctx modules.Context) error {
upstreamURL := strings.TrimSpace(ctx.Config.AmpUpstreamURL)
// Determine auth middleware (from module or context)
auth := m.getAuthMiddleware(ctx)
// Use registerOnce to ensure routes are only registered once
var regErr error
m.registerOnce.Do(func() {
// Always register provider aliases - these work without an upstream
m.registerProviderAliases(ctx.Engine, ctx.BaseHandler, auth)
// If no upstream URL, skip proxy routes but provider aliases are still available
if upstreamURL == "" {
log.Debug("Amp upstream proxy disabled (no upstream URL configured)")
log.Debug("Amp provider alias routes registered")
m.enabled = false
return
}
// Create secret source with precedence: config > env > file
// Cache secrets for 5 minutes to reduce file I/O
if m.secretSource == nil {
m.secretSource = NewMultiSourceSecret(ctx.Config.AmpUpstreamAPIKey, 0 /* default 5min */)
}
// Create reverse proxy with gzip handling via ModifyResponse
proxy, err := createReverseProxy(upstreamURL, m.secretSource)
if err != nil {
regErr = fmt.Errorf("failed to create amp proxy: %w", err)
return
}
m.proxy = proxy
m.enabled = true
// Register management proxy routes (requires upstream)
// Restrict to localhost by default for security (prevents drive-by browser attacks)
handler := proxyHandler(proxy)
m.registerManagementRoutes(ctx.Engine, ctx.BaseHandler, handler, ctx.Config.AmpRestrictManagementToLocalhost)
log.Infof("Amp upstream proxy enabled for: %s", upstreamURL)
log.Debug("Amp provider alias routes registered")
})
return regErr
}
// getAuthMiddleware returns the authentication middleware, preferring the
// module's configured middleware, then the context middleware, then a fallback.
func (m *AmpModule) getAuthMiddleware(ctx modules.Context) gin.HandlerFunc {
if m.authMiddleware_ != nil {
return m.authMiddleware_
}
if ctx.AuthMiddleware != nil {
return ctx.AuthMiddleware
}
// Fallback: no authentication (should not happen in production)
log.Warn("Amp module: no auth middleware provided, allowing all requests")
return func(c *gin.Context) {
c.Next()
}
}
// OnConfigUpdated handles configuration updates.
// Currently requires restart for URL changes (could be enhanced for dynamic updates).
func (m *AmpModule) OnConfigUpdated(cfg *config.Config) error {
if !m.enabled {
log.Debug("Amp routing not enabled, skipping config update")
return nil
}
upstreamURL := strings.TrimSpace(cfg.AmpUpstreamURL)
if upstreamURL == "" {
log.Warn("Amp upstream URL removed from config, restart required to disable")
return nil
}
// If API key changed, invalidate the cache
if m.secretSource != nil {
if ms, ok := m.secretSource.(*MultiSourceSecret); ok {
ms.InvalidateCache()
log.Debug("Amp secret cache invalidated due to config update")
}
}
log.Debug("Amp config updated (restart required for URL changes)")
return nil
}

View File

@@ -0,0 +1,303 @@
package amp
import (
"context"
"net/http/httptest"
"os"
"path/filepath"
"testing"
"time"
"github.com/gin-gonic/gin"
"github.com/router-for-me/CLIProxyAPI/v6/internal/api/modules"
"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
sdkaccess "github.com/router-for-me/CLIProxyAPI/v6/sdk/access"
"github.com/router-for-me/CLIProxyAPI/v6/sdk/api/handlers"
)
func TestAmpModule_Name(t *testing.T) {
m := New()
if m.Name() != "amp-routing" {
t.Fatalf("want amp-routing, got %s", m.Name())
}
}
func TestAmpModule_New(t *testing.T) {
accessManager := sdkaccess.NewManager()
authMiddleware := func(c *gin.Context) { c.Next() }
m := NewLegacy(accessManager, authMiddleware)
if m.accessManager != accessManager {
t.Fatal("accessManager not set")
}
if m.authMiddleware_ == nil {
t.Fatal("authMiddleware not set")
}
if m.enabled {
t.Fatal("enabled should be false initially")
}
if m.proxy != nil {
t.Fatal("proxy should be nil initially")
}
}
func TestAmpModule_Register_WithUpstream(t *testing.T) {
gin.SetMode(gin.TestMode)
r := gin.New()
// Fake upstream to ensure URL is valid
upstream := httptest.NewServer(nil)
defer upstream.Close()
accessManager := sdkaccess.NewManager()
base := &handlers.BaseAPIHandler{}
m := NewLegacy(accessManager, func(c *gin.Context) { c.Next() })
cfg := &config.Config{
AmpUpstreamURL: upstream.URL,
AmpUpstreamAPIKey: "test-key",
}
ctx := modules.Context{Engine: r, BaseHandler: base, Config: cfg, AuthMiddleware: func(c *gin.Context) { c.Next() }}
if err := m.Register(ctx); err != nil {
t.Fatalf("register error: %v", err)
}
if !m.enabled {
t.Fatal("module should be enabled with upstream URL")
}
if m.proxy == nil {
t.Fatal("proxy should be initialized")
}
if m.secretSource == nil {
t.Fatal("secretSource should be initialized")
}
}
func TestAmpModule_Register_WithoutUpstream(t *testing.T) {
gin.SetMode(gin.TestMode)
r := gin.New()
accessManager := sdkaccess.NewManager()
base := &handlers.BaseAPIHandler{}
m := NewLegacy(accessManager, func(c *gin.Context) { c.Next() })
cfg := &config.Config{
AmpUpstreamURL: "", // No upstream
}
ctx := modules.Context{Engine: r, BaseHandler: base, Config: cfg, AuthMiddleware: func(c *gin.Context) { c.Next() }}
if err := m.Register(ctx); err != nil {
t.Fatalf("register should not error without upstream: %v", err)
}
if m.enabled {
t.Fatal("module should be disabled without upstream URL")
}
if m.proxy != nil {
t.Fatal("proxy should not be initialized without upstream")
}
// But provider aliases should still be registered
req := httptest.NewRequest("GET", "/api/provider/openai/models", nil)
w := httptest.NewRecorder()
r.ServeHTTP(w, req)
if w.Code == 404 {
t.Fatal("provider aliases should be registered even without upstream")
}
}
func TestAmpModule_Register_InvalidUpstream(t *testing.T) {
gin.SetMode(gin.TestMode)
r := gin.New()
accessManager := sdkaccess.NewManager()
base := &handlers.BaseAPIHandler{}
m := NewLegacy(accessManager, func(c *gin.Context) { c.Next() })
cfg := &config.Config{
AmpUpstreamURL: "://invalid-url",
}
ctx := modules.Context{Engine: r, BaseHandler: base, Config: cfg, AuthMiddleware: func(c *gin.Context) { c.Next() }}
if err := m.Register(ctx); err == nil {
t.Fatal("expected error for invalid upstream URL")
}
}
func TestAmpModule_OnConfigUpdated_CacheInvalidation(t *testing.T) {
tmpDir := t.TempDir()
p := filepath.Join(tmpDir, "secrets.json")
if err := os.WriteFile(p, []byte(`{"apiKey@https://ampcode.com/":"v1"}`), 0600); err != nil {
t.Fatal(err)
}
m := &AmpModule{enabled: true}
ms := NewMultiSourceSecretWithPath("", p, time.Minute)
m.secretSource = ms
// Warm the cache
if _, err := ms.Get(context.Background()); err != nil {
t.Fatal(err)
}
if ms.cache == nil {
t.Fatal("expected cache to be set")
}
// Update config - should invalidate cache
if err := m.OnConfigUpdated(&config.Config{AmpUpstreamURL: "http://x"}); err != nil {
t.Fatal(err)
}
if ms.cache != nil {
t.Fatal("expected cache to be invalidated")
}
}
func TestAmpModule_OnConfigUpdated_NotEnabled(t *testing.T) {
m := &AmpModule{enabled: false}
// Should not error or panic when disabled
if err := m.OnConfigUpdated(&config.Config{}); err != nil {
t.Fatalf("unexpected error: %v", err)
}
}
func TestAmpModule_OnConfigUpdated_URLRemoved(t *testing.T) {
m := &AmpModule{enabled: true}
ms := NewMultiSourceSecret("", 0)
m.secretSource = ms
// Config update with empty URL - should log warning but not error
cfg := &config.Config{AmpUpstreamURL: ""}
if err := m.OnConfigUpdated(cfg); err != nil {
t.Fatalf("unexpected error: %v", err)
}
}
func TestAmpModule_OnConfigUpdated_NonMultiSourceSecret(t *testing.T) {
// Test that OnConfigUpdated doesn't panic with StaticSecretSource
m := &AmpModule{enabled: true}
m.secretSource = NewStaticSecretSource("static-key")
cfg := &config.Config{AmpUpstreamURL: "http://example.com"}
// Should not error or panic
if err := m.OnConfigUpdated(cfg); err != nil {
t.Fatalf("unexpected error: %v", err)
}
}
func TestAmpModule_AuthMiddleware_Fallback(t *testing.T) {
gin.SetMode(gin.TestMode)
r := gin.New()
// Create module with no auth middleware
m := &AmpModule{authMiddleware_: nil}
// Get the fallback middleware via getAuthMiddleware
ctx := modules.Context{Engine: r, AuthMiddleware: nil}
middleware := m.getAuthMiddleware(ctx)
if middleware == nil {
t.Fatal("getAuthMiddleware should return a fallback, not nil")
}
// Test that it works
called := false
r.GET("/test", middleware, func(c *gin.Context) {
called = true
c.String(200, "ok")
})
req := httptest.NewRequest("GET", "/test", nil)
w := httptest.NewRecorder()
r.ServeHTTP(w, req)
if !called {
t.Fatal("fallback middleware should allow requests through")
}
}
func TestAmpModule_SecretSource_FromConfig(t *testing.T) {
gin.SetMode(gin.TestMode)
r := gin.New()
upstream := httptest.NewServer(nil)
defer upstream.Close()
accessManager := sdkaccess.NewManager()
base := &handlers.BaseAPIHandler{}
m := NewLegacy(accessManager, func(c *gin.Context) { c.Next() })
// Config with explicit API key
cfg := &config.Config{
AmpUpstreamURL: upstream.URL,
AmpUpstreamAPIKey: "config-key",
}
ctx := modules.Context{Engine: r, BaseHandler: base, Config: cfg, AuthMiddleware: func(c *gin.Context) { c.Next() }}
if err := m.Register(ctx); err != nil {
t.Fatalf("register error: %v", err)
}
// Secret source should be MultiSourceSecret with config key
if m.secretSource == nil {
t.Fatal("secretSource should be set")
}
// Verify it returns the config key
key, err := m.secretSource.Get(context.Background())
if err != nil {
t.Fatalf("Get error: %v", err)
}
if key != "config-key" {
t.Fatalf("want config-key, got %s", key)
}
}
func TestAmpModule_ProviderAliasesAlwaysRegistered(t *testing.T) {
gin.SetMode(gin.TestMode)
scenarios := []struct {
name string
configURL string
}{
{"with_upstream", "http://example.com"},
{"without_upstream", ""},
}
for _, scenario := range scenarios {
t.Run(scenario.name, func(t *testing.T) {
r := gin.New()
accessManager := sdkaccess.NewManager()
base := &handlers.BaseAPIHandler{}
m := NewLegacy(accessManager, func(c *gin.Context) { c.Next() })
cfg := &config.Config{AmpUpstreamURL: scenario.configURL}
ctx := modules.Context{Engine: r, BaseHandler: base, Config: cfg, AuthMiddleware: func(c *gin.Context) { c.Next() }}
if err := m.Register(ctx); err != nil && scenario.configURL != "" {
t.Fatalf("register error: %v", err)
}
// Provider aliases should always be available
req := httptest.NewRequest("GET", "/api/provider/openai/models", nil)
w := httptest.NewRecorder()
r.ServeHTTP(w, req)
if w.Code == 404 {
t.Fatal("provider aliases should be registered")
}
})
}
}

View File

@@ -0,0 +1,129 @@
package amp
import (
"bytes"
"encoding/json"
"io"
"net/http/httputil"
"strings"
"github.com/gin-gonic/gin"
"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
log "github.com/sirupsen/logrus"
)
// FallbackHandler wraps a standard handler with fallback logic to ampcode.com
// when the model's provider is not available in CLIProxyAPI
type FallbackHandler struct {
getProxy func() *httputil.ReverseProxy
}
// NewFallbackHandler creates a new fallback handler wrapper
// The getProxy function allows lazy evaluation of the proxy (useful when proxy is created after routes)
func NewFallbackHandler(getProxy func() *httputil.ReverseProxy) *FallbackHandler {
return &FallbackHandler{
getProxy: getProxy,
}
}
// WrapHandler wraps a gin.HandlerFunc with fallback logic
// If the model's provider is not configured in CLIProxyAPI, it forwards to ampcode.com
func (fh *FallbackHandler) WrapHandler(handler gin.HandlerFunc) gin.HandlerFunc {
return func(c *gin.Context) {
// Read the request body to extract the model name
bodyBytes, err := io.ReadAll(c.Request.Body)
if err != nil {
log.Errorf("amp fallback: failed to read request body: %v", err)
handler(c)
return
}
// Restore the body for the handler to read
c.Request.Body = io.NopCloser(bytes.NewReader(bodyBytes))
// Try to extract model from request body or URL path (for Gemini)
modelName := extractModelFromRequest(bodyBytes, c)
if modelName == "" {
// Can't determine model, proceed with normal handler
handler(c)
return
}
// Normalize model (handles Gemini thinking suffixes)
normalizedModel, _ := util.NormalizeGeminiThinkingModel(modelName)
// Check if we have providers for this model
providers := util.GetProviderName(normalizedModel)
if len(providers) == 0 {
// No providers configured - check if we have a proxy for fallback
proxy := fh.getProxy()
if proxy != nil {
// Fallback to ampcode.com
log.Infof("amp fallback: model %s has no configured provider, forwarding to ampcode.com", modelName)
// Restore body again for the proxy
c.Request.Body = io.NopCloser(bytes.NewReader(bodyBytes))
// Forward to ampcode.com
proxy.ServeHTTP(c.Writer, c.Request)
return
}
// No proxy available, let the normal handler return the error
log.Debugf("amp fallback: model %s has no configured provider and no proxy available", modelName)
}
// Providers available or no proxy for fallback, restore body and use normal handler
// Filter Anthropic-Beta header to remove features requiring special subscription
// This is needed when using local providers (bypassing the Amp proxy)
if betaHeader := c.Request.Header.Get("Anthropic-Beta"); betaHeader != "" {
filtered := filterBetaFeatures(betaHeader, "context-1m-2025-08-07")
if filtered != "" {
c.Request.Header.Set("Anthropic-Beta", filtered)
} else {
c.Request.Header.Del("Anthropic-Beta")
}
}
c.Request.Body = io.NopCloser(bytes.NewReader(bodyBytes))
handler(c)
}
}
// extractModelFromRequest attempts to extract the model name from various request formats
func extractModelFromRequest(body []byte, c *gin.Context) string {
// First try to parse from JSON body (OpenAI, Claude, etc.)
var payload map[string]interface{}
if err := json.Unmarshal(body, &payload); err == nil {
// Check common model field names
if model, ok := payload["model"].(string); ok {
return model
}
}
// For Gemini requests, model is in the URL path
// Standard format: /models/{model}:generateContent -> :action parameter
if action := c.Param("action"); action != "" {
// Split by colon to get model name (e.g., "gemini-pro:generateContent" -> "gemini-pro")
parts := strings.Split(action, ":")
if len(parts) > 0 && parts[0] != "" {
return parts[0]
}
}
// AMP CLI format: /publishers/google/models/{model}:method -> *path parameter
// Example: /publishers/google/models/gemini-3-pro-preview:streamGenerateContent
if path := c.Param("path"); path != "" {
// Look for /models/{model}:method pattern
if idx := strings.Index(path, "/models/"); idx >= 0 {
modelPart := path[idx+8:] // Skip "/models/"
// Split by colon to get model name
if colonIdx := strings.Index(modelPart, ":"); colonIdx > 0 {
return modelPart[:colonIdx]
}
}
}
return ""
}

View File

@@ -0,0 +1,45 @@
package amp
import (
"strings"
"github.com/gin-gonic/gin"
"github.com/router-for-me/CLIProxyAPI/v6/sdk/api/handlers/gemini"
)
// createGeminiBridgeHandler creates a handler that bridges AMP CLI's non-standard Gemini paths
// to our standard Gemini handler by rewriting the request context.
//
// AMP CLI format: /publishers/google/models/gemini-3-pro-preview:streamGenerateContent
// Standard format: /models/gemini-3-pro-preview:streamGenerateContent
//
// This extracts the model+method from the AMP path and sets it as the :action parameter
// so the standard Gemini handler can process it.
func createGeminiBridgeHandler(geminiHandler *gemini.GeminiAPIHandler) gin.HandlerFunc {
return func(c *gin.Context) {
// Get the full path from the catch-all parameter
path := c.Param("path")
// Extract model:method from AMP CLI path format
// Example: /publishers/google/models/gemini-3-pro-preview:streamGenerateContent
if idx := strings.Index(path, "/models/"); idx >= 0 {
// Extract everything after "/models/"
actionPart := path[idx+8:] // Skip "/models/"
// Set this as the :action parameter that the Gemini handler expects
c.Params = append(c.Params, gin.Param{
Key: "action",
Value: actionPart,
})
// Call the standard Gemini handler
geminiHandler.GeminiHandler(c)
return
}
// If we can't parse the path, return 400
c.JSON(400, gin.H{
"error": "Invalid Gemini API path format",
})
}
}

View File

@@ -0,0 +1,195 @@
package amp
import (
"bytes"
"compress/gzip"
"fmt"
"io"
"net/http"
"net/http/httputil"
"net/url"
"strconv"
"strings"
"github.com/gin-gonic/gin"
log "github.com/sirupsen/logrus"
)
// readCloser wraps a reader and forwards Close to a separate closer.
// Used to restore peeked bytes while preserving upstream body Close behavior.
type readCloser struct {
r io.Reader
c io.Closer
}
func (rc *readCloser) Read(p []byte) (int, error) { return rc.r.Read(p) }
func (rc *readCloser) Close() error { return rc.c.Close() }
// createReverseProxy creates a reverse proxy handler for Amp upstream
// with automatic gzip decompression via ModifyResponse
func createReverseProxy(upstreamURL string, secretSource SecretSource) (*httputil.ReverseProxy, error) {
parsed, err := url.Parse(upstreamURL)
if err != nil {
return nil, fmt.Errorf("invalid amp upstream url: %w", err)
}
proxy := httputil.NewSingleHostReverseProxy(parsed)
originalDirector := proxy.Director
// Modify outgoing requests to inject API key and fix routing
proxy.Director = func(req *http.Request) {
originalDirector(req)
req.Host = parsed.Host
// Preserve correlation headers for debugging
if req.Header.Get("X-Request-ID") == "" {
// Could generate one here if needed
}
// Note: We do NOT filter Anthropic-Beta headers in the proxy path
// Users going through ampcode.com proxy are paying for the service and should get all features
// including 1M context window (context-1m-2025-08-07)
// Inject API key from secret source (precedence: config > env > file)
if key, err := secretSource.Get(req.Context()); err == nil && key != "" {
req.Header.Set("X-Api-Key", key)
req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", key))
} else if err != nil {
log.Warnf("amp secret source error (continuing without auth): %v", err)
}
}
// Modify incoming responses to handle gzip without Content-Encoding
// This addresses the same issue as inline handler gzip handling, but at the proxy level
proxy.ModifyResponse = func(resp *http.Response) error {
// Only process successful responses
if resp.StatusCode < 200 || resp.StatusCode >= 300 {
return nil
}
// Skip if already marked as gzip (Content-Encoding set)
if resp.Header.Get("Content-Encoding") != "" {
return nil
}
// Skip streaming responses (SSE, chunked)
if isStreamingResponse(resp) {
return nil
}
// Save reference to original upstream body for proper cleanup
originalBody := resp.Body
// Peek at first 2 bytes to detect gzip magic bytes
header := make([]byte, 2)
n, _ := io.ReadFull(originalBody, header)
// Check for gzip magic bytes (0x1f 0x8b)
// If n < 2, we didn't get enough bytes, so it's not gzip
if n >= 2 && header[0] == 0x1f && header[1] == 0x8b {
// It's gzip - read the rest of the body
rest, err := io.ReadAll(originalBody)
if err != nil {
// Restore what we read and return original body (preserve Close behavior)
resp.Body = &readCloser{
r: io.MultiReader(bytes.NewReader(header[:n]), originalBody),
c: originalBody,
}
return nil
}
// Reconstruct complete gzipped data
gzippedData := append(header[:n], rest...)
// Decompress
gzipReader, err := gzip.NewReader(bytes.NewReader(gzippedData))
if err != nil {
log.Warnf("amp proxy: gzip header detected but decompress failed: %v", err)
// Close original body and return in-memory copy
_ = originalBody.Close()
resp.Body = io.NopCloser(bytes.NewReader(gzippedData))
return nil
}
decompressed, err := io.ReadAll(gzipReader)
_ = gzipReader.Close()
if err != nil {
log.Warnf("amp proxy: gzip decompress error: %v", err)
// Close original body and return in-memory copy
_ = originalBody.Close()
resp.Body = io.NopCloser(bytes.NewReader(gzippedData))
return nil
}
// Close original body since we're replacing with in-memory decompressed content
_ = originalBody.Close()
// Replace body with decompressed content
resp.Body = io.NopCloser(bytes.NewReader(decompressed))
resp.ContentLength = int64(len(decompressed))
// Update headers to reflect decompressed state
resp.Header.Del("Content-Encoding") // No longer compressed
resp.Header.Del("Content-Length") // Remove stale compressed length
resp.Header.Set("Content-Length", strconv.FormatInt(resp.ContentLength, 10)) // Set decompressed length
log.Debugf("amp proxy: decompressed gzip response (%d -> %d bytes)", len(gzippedData), len(decompressed))
} else {
// Not gzip - restore peeked bytes while preserving Close behavior
// Handle edge cases: n might be 0, 1, or 2 depending on EOF
resp.Body = &readCloser{
r: io.MultiReader(bytes.NewReader(header[:n]), originalBody),
c: originalBody,
}
}
return nil
}
// Error handler for proxy failures
proxy.ErrorHandler = func(rw http.ResponseWriter, req *http.Request, err error) {
log.Errorf("amp upstream proxy error for %s %s: %v", req.Method, req.URL.Path, err)
rw.Header().Set("Content-Type", "application/json")
rw.WriteHeader(http.StatusBadGateway)
_, _ = rw.Write([]byte(`{"error":"amp_upstream_proxy_error","message":"Failed to reach Amp upstream"}`))
}
return proxy, nil
}
// isStreamingResponse detects if the response is streaming (SSE only)
// Note: We only treat text/event-stream as streaming. Chunked transfer encoding
// is a transport-level detail and doesn't mean we can't decompress the full response.
// Many JSON APIs use chunked encoding for normal responses.
func isStreamingResponse(resp *http.Response) bool {
contentType := resp.Header.Get("Content-Type")
// Only Server-Sent Events are true streaming responses
if strings.Contains(contentType, "text/event-stream") {
return true
}
return false
}
// proxyHandler converts httputil.ReverseProxy to gin.HandlerFunc
func proxyHandler(proxy *httputil.ReverseProxy) gin.HandlerFunc {
return func(c *gin.Context) {
proxy.ServeHTTP(c.Writer, c.Request)
}
}
// filterBetaFeatures removes a specific beta feature from comma-separated list
func filterBetaFeatures(header, featureToRemove string) string {
features := strings.Split(header, ",")
filtered := make([]string, 0, len(features))
for _, feature := range features {
trimmed := strings.TrimSpace(feature)
if trimmed != "" && trimmed != featureToRemove {
filtered = append(filtered, trimmed)
}
}
return strings.Join(filtered, ",")
}

View File

@@ -0,0 +1,500 @@
package amp
import (
"bytes"
"compress/gzip"
"fmt"
"io"
"net/http"
"net/http/httptest"
"testing"
)
// Helper: compress data with gzip
func gzipBytes(b []byte) []byte {
var buf bytes.Buffer
zw := gzip.NewWriter(&buf)
zw.Write(b)
zw.Close()
return buf.Bytes()
}
// Helper: create a mock http.Response
func mkResp(status int, hdr http.Header, body []byte) *http.Response {
if hdr == nil {
hdr = http.Header{}
}
return &http.Response{
StatusCode: status,
Header: hdr,
Body: io.NopCloser(bytes.NewReader(body)),
ContentLength: int64(len(body)),
}
}
func TestCreateReverseProxy_ValidURL(t *testing.T) {
proxy, err := createReverseProxy("http://example.com", NewStaticSecretSource("key"))
if err != nil {
t.Fatalf("expected no error, got: %v", err)
}
if proxy == nil {
t.Fatal("expected proxy to be created")
}
}
func TestCreateReverseProxy_InvalidURL(t *testing.T) {
_, err := createReverseProxy("://invalid", NewStaticSecretSource("key"))
if err == nil {
t.Fatal("expected error for invalid URL")
}
}
func TestModifyResponse_GzipScenarios(t *testing.T) {
proxy, err := createReverseProxy("http://example.com", NewStaticSecretSource("k"))
if err != nil {
t.Fatal(err)
}
goodJSON := []byte(`{"ok":true}`)
good := gzipBytes(goodJSON)
truncated := good[:10]
corrupted := append([]byte{0x1f, 0x8b}, []byte("notgzip")...)
cases := []struct {
name string
header http.Header
body []byte
status int
wantBody []byte
wantCE string
}{
{
name: "decompresses_valid_gzip_no_header",
header: http.Header{},
body: good,
status: 200,
wantBody: goodJSON,
wantCE: "",
},
{
name: "skips_when_ce_present",
header: http.Header{"Content-Encoding": []string{"gzip"}},
body: good,
status: 200,
wantBody: good,
wantCE: "gzip",
},
{
name: "passes_truncated_unchanged",
header: http.Header{},
body: truncated,
status: 200,
wantBody: truncated,
wantCE: "",
},
{
name: "passes_corrupted_unchanged",
header: http.Header{},
body: corrupted,
status: 200,
wantBody: corrupted,
wantCE: "",
},
{
name: "non_gzip_unchanged",
header: http.Header{},
body: []byte("plain"),
status: 200,
wantBody: []byte("plain"),
wantCE: "",
},
{
name: "empty_body",
header: http.Header{},
body: []byte{},
status: 200,
wantBody: []byte{},
wantCE: "",
},
{
name: "single_byte_body",
header: http.Header{},
body: []byte{0x1f},
status: 200,
wantBody: []byte{0x1f},
wantCE: "",
},
{
name: "skips_non_2xx_status",
header: http.Header{},
body: good,
status: 404,
wantBody: good,
wantCE: "",
},
}
for _, tc := range cases {
t.Run(tc.name, func(t *testing.T) {
resp := mkResp(tc.status, tc.header, tc.body)
if err := proxy.ModifyResponse(resp); err != nil {
t.Fatalf("ModifyResponse error: %v", err)
}
got, err := io.ReadAll(resp.Body)
if err != nil {
t.Fatalf("ReadAll error: %v", err)
}
if !bytes.Equal(got, tc.wantBody) {
t.Fatalf("body mismatch:\nwant: %q\ngot: %q", tc.wantBody, got)
}
if ce := resp.Header.Get("Content-Encoding"); ce != tc.wantCE {
t.Fatalf("Content-Encoding: want %q, got %q", tc.wantCE, ce)
}
})
}
}
func TestModifyResponse_UpdatesContentLengthHeader(t *testing.T) {
proxy, err := createReverseProxy("http://example.com", NewStaticSecretSource("k"))
if err != nil {
t.Fatal(err)
}
goodJSON := []byte(`{"message":"test response"}`)
gzipped := gzipBytes(goodJSON)
// Simulate upstream response with gzip body AND Content-Length header
// (this is the scenario the bot flagged - stale Content-Length after decompression)
resp := mkResp(200, http.Header{
"Content-Length": []string{fmt.Sprintf("%d", len(gzipped))}, // Compressed size
}, gzipped)
if err := proxy.ModifyResponse(resp); err != nil {
t.Fatalf("ModifyResponse error: %v", err)
}
// Verify body is decompressed
got, _ := io.ReadAll(resp.Body)
if !bytes.Equal(got, goodJSON) {
t.Fatalf("body should be decompressed, got: %q, want: %q", got, goodJSON)
}
// Verify Content-Length header is updated to decompressed size
wantCL := fmt.Sprintf("%d", len(goodJSON))
gotCL := resp.Header.Get("Content-Length")
if gotCL != wantCL {
t.Fatalf("Content-Length header mismatch: want %q (decompressed), got %q", wantCL, gotCL)
}
// Verify struct field also matches
if resp.ContentLength != int64(len(goodJSON)) {
t.Fatalf("resp.ContentLength mismatch: want %d, got %d", len(goodJSON), resp.ContentLength)
}
}
func TestModifyResponse_SkipsStreamingResponses(t *testing.T) {
proxy, err := createReverseProxy("http://example.com", NewStaticSecretSource("k"))
if err != nil {
t.Fatal(err)
}
goodJSON := []byte(`{"ok":true}`)
gzipped := gzipBytes(goodJSON)
t.Run("sse_skips_decompression", func(t *testing.T) {
resp := mkResp(200, http.Header{"Content-Type": []string{"text/event-stream"}}, gzipped)
if err := proxy.ModifyResponse(resp); err != nil {
t.Fatalf("ModifyResponse error: %v", err)
}
// SSE should NOT be decompressed
got, _ := io.ReadAll(resp.Body)
if !bytes.Equal(got, gzipped) {
t.Fatal("SSE response should not be decompressed")
}
})
}
func TestModifyResponse_DecompressesChunkedJSON(t *testing.T) {
proxy, err := createReverseProxy("http://example.com", NewStaticSecretSource("k"))
if err != nil {
t.Fatal(err)
}
goodJSON := []byte(`{"ok":true}`)
gzipped := gzipBytes(goodJSON)
t.Run("chunked_json_decompresses", func(t *testing.T) {
// Chunked JSON responses (like thread APIs) should be decompressed
resp := mkResp(200, http.Header{"Transfer-Encoding": []string{"chunked"}}, gzipped)
if err := proxy.ModifyResponse(resp); err != nil {
t.Fatalf("ModifyResponse error: %v", err)
}
// Should decompress because it's not SSE
got, _ := io.ReadAll(resp.Body)
if !bytes.Equal(got, goodJSON) {
t.Fatalf("chunked JSON should be decompressed, got: %q, want: %q", got, goodJSON)
}
})
}
func TestReverseProxy_InjectsHeaders(t *testing.T) {
gotHeaders := make(chan http.Header, 1)
upstream := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
gotHeaders <- r.Header.Clone()
w.WriteHeader(200)
w.Write([]byte(`ok`))
}))
defer upstream.Close()
proxy, err := createReverseProxy(upstream.URL, NewStaticSecretSource("secret"))
if err != nil {
t.Fatal(err)
}
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
proxy.ServeHTTP(w, r)
}))
defer srv.Close()
res, err := http.Get(srv.URL + "/test")
if err != nil {
t.Fatal(err)
}
res.Body.Close()
hdr := <-gotHeaders
if hdr.Get("X-Api-Key") != "secret" {
t.Fatalf("X-Api-Key missing or wrong, got: %q", hdr.Get("X-Api-Key"))
}
if hdr.Get("Authorization") != "Bearer secret" {
t.Fatalf("Authorization missing or wrong, got: %q", hdr.Get("Authorization"))
}
}
func TestReverseProxy_EmptySecret(t *testing.T) {
gotHeaders := make(chan http.Header, 1)
upstream := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
gotHeaders <- r.Header.Clone()
w.WriteHeader(200)
w.Write([]byte(`ok`))
}))
defer upstream.Close()
proxy, err := createReverseProxy(upstream.URL, NewStaticSecretSource(""))
if err != nil {
t.Fatal(err)
}
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
proxy.ServeHTTP(w, r)
}))
defer srv.Close()
res, err := http.Get(srv.URL + "/test")
if err != nil {
t.Fatal(err)
}
res.Body.Close()
hdr := <-gotHeaders
// Should NOT inject headers when secret is empty
if hdr.Get("X-Api-Key") != "" {
t.Fatalf("X-Api-Key should not be set, got: %q", hdr.Get("X-Api-Key"))
}
if authVal := hdr.Get("Authorization"); authVal != "" && authVal != "Bearer " {
t.Fatalf("Authorization should not be set, got: %q", authVal)
}
}
func TestReverseProxy_ErrorHandler(t *testing.T) {
// Point proxy to a non-routable address to trigger error
proxy, err := createReverseProxy("http://127.0.0.1:1", NewStaticSecretSource(""))
if err != nil {
t.Fatal(err)
}
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
proxy.ServeHTTP(w, r)
}))
defer srv.Close()
res, err := http.Get(srv.URL + "/any")
if err != nil {
t.Fatal(err)
}
body, _ := io.ReadAll(res.Body)
res.Body.Close()
if res.StatusCode != http.StatusBadGateway {
t.Fatalf("want 502, got %d", res.StatusCode)
}
if !bytes.Contains(body, []byte(`"amp_upstream_proxy_error"`)) {
t.Fatalf("unexpected body: %s", body)
}
if ct := res.Header.Get("Content-Type"); ct != "application/json" {
t.Fatalf("content-type: want application/json, got %s", ct)
}
}
func TestReverseProxy_FullRoundTrip_Gzip(t *testing.T) {
// Upstream returns gzipped JSON without Content-Encoding header
upstream := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
w.WriteHeader(200)
w.Write(gzipBytes([]byte(`{"upstream":"ok"}`)))
}))
defer upstream.Close()
proxy, err := createReverseProxy(upstream.URL, NewStaticSecretSource("key"))
if err != nil {
t.Fatal(err)
}
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
proxy.ServeHTTP(w, r)
}))
defer srv.Close()
res, err := http.Get(srv.URL + "/test")
if err != nil {
t.Fatal(err)
}
body, _ := io.ReadAll(res.Body)
res.Body.Close()
expected := []byte(`{"upstream":"ok"}`)
if !bytes.Equal(body, expected) {
t.Fatalf("want decompressed JSON, got: %s", body)
}
}
func TestReverseProxy_FullRoundTrip_PlainJSON(t *testing.T) {
// Upstream returns plain JSON
upstream := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
w.Header().Set("Content-Type", "application/json")
w.WriteHeader(200)
w.Write([]byte(`{"plain":"json"}`))
}))
defer upstream.Close()
proxy, err := createReverseProxy(upstream.URL, NewStaticSecretSource("key"))
if err != nil {
t.Fatal(err)
}
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
proxy.ServeHTTP(w, r)
}))
defer srv.Close()
res, err := http.Get(srv.URL + "/test")
if err != nil {
t.Fatal(err)
}
body, _ := io.ReadAll(res.Body)
res.Body.Close()
expected := []byte(`{"plain":"json"}`)
if !bytes.Equal(body, expected) {
t.Fatalf("want plain JSON unchanged, got: %s", body)
}
}
func TestIsStreamingResponse(t *testing.T) {
cases := []struct {
name string
header http.Header
want bool
}{
{
name: "sse",
header: http.Header{"Content-Type": []string{"text/event-stream"}},
want: true,
},
{
name: "chunked_not_streaming",
header: http.Header{"Transfer-Encoding": []string{"chunked"}},
want: false, // Chunked is transport-level, not streaming
},
{
name: "normal_json",
header: http.Header{"Content-Type": []string{"application/json"}},
want: false,
},
{
name: "empty",
header: http.Header{},
want: false,
},
}
for _, tc := range cases {
t.Run(tc.name, func(t *testing.T) {
resp := &http.Response{Header: tc.header}
got := isStreamingResponse(resp)
if got != tc.want {
t.Fatalf("want %v, got %v", tc.want, got)
}
})
}
}
func TestFilterBetaFeatures(t *testing.T) {
tests := []struct {
name string
header string
featureToRemove string
expected string
}{
{
name: "Remove context-1m from middle",
header: "fine-grained-tool-streaming-2025-05-14,context-1m-2025-08-07,oauth-2025-04-20",
featureToRemove: "context-1m-2025-08-07",
expected: "fine-grained-tool-streaming-2025-05-14,oauth-2025-04-20",
},
{
name: "Remove context-1m from start",
header: "context-1m-2025-08-07,fine-grained-tool-streaming-2025-05-14",
featureToRemove: "context-1m-2025-08-07",
expected: "fine-grained-tool-streaming-2025-05-14",
},
{
name: "Remove context-1m from end",
header: "fine-grained-tool-streaming-2025-05-14,context-1m-2025-08-07",
featureToRemove: "context-1m-2025-08-07",
expected: "fine-grained-tool-streaming-2025-05-14",
},
{
name: "Feature not present",
header: "fine-grained-tool-streaming-2025-05-14,oauth-2025-04-20",
featureToRemove: "context-1m-2025-08-07",
expected: "fine-grained-tool-streaming-2025-05-14,oauth-2025-04-20",
},
{
name: "Only feature to remove",
header: "context-1m-2025-08-07",
featureToRemove: "context-1m-2025-08-07",
expected: "",
},
{
name: "Empty header",
header: "",
featureToRemove: "context-1m-2025-08-07",
expected: "",
},
{
name: "Header with spaces",
header: "fine-grained-tool-streaming-2025-05-14, context-1m-2025-08-07 , oauth-2025-04-20",
featureToRemove: "context-1m-2025-08-07",
expected: "fine-grained-tool-streaming-2025-05-14,oauth-2025-04-20",
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
result := filterBetaFeatures(tt.header, tt.featureToRemove)
if result != tt.expected {
t.Errorf("filterBetaFeatures() = %q, want %q", result, tt.expected)
}
})
}
}

View File

@@ -0,0 +1,222 @@
package amp
import (
"net"
"net/http/httputil"
"strings"
"github.com/gin-gonic/gin"
"github.com/router-for-me/CLIProxyAPI/v6/sdk/api/handlers"
"github.com/router-for-me/CLIProxyAPI/v6/sdk/api/handlers/claude"
"github.com/router-for-me/CLIProxyAPI/v6/sdk/api/handlers/gemini"
"github.com/router-for-me/CLIProxyAPI/v6/sdk/api/handlers/openai"
"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
log "github.com/sirupsen/logrus"
)
// localhostOnlyMiddleware restricts access to localhost (127.0.0.1, ::1) only.
// Returns 403 Forbidden for non-localhost clients.
//
// Security: Uses RemoteAddr (actual TCP connection) instead of ClientIP() to prevent
// header spoofing attacks via X-Forwarded-For or similar headers. This means the
// middleware will not work correctly behind reverse proxies - users deploying behind
// nginx/Cloudflare should disable this feature and use firewall rules instead.
func localhostOnlyMiddleware() gin.HandlerFunc {
return func(c *gin.Context) {
// Use actual TCP connection address (RemoteAddr) to prevent header spoofing
// This cannot be forged by X-Forwarded-For or other client-controlled headers
remoteAddr := c.Request.RemoteAddr
// RemoteAddr format is "IP:port" or "[IPv6]:port", extract just the IP
host, _, err := net.SplitHostPort(remoteAddr)
if err != nil {
// Try parsing as raw IP (shouldn't happen with standard HTTP, but be defensive)
host = remoteAddr
}
// Parse the IP to handle both IPv4 and IPv6
ip := net.ParseIP(host)
if ip == nil {
log.Warnf("Amp management: invalid RemoteAddr %s, denying access", remoteAddr)
c.AbortWithStatusJSON(403, gin.H{
"error": "Access denied: management routes restricted to localhost",
})
return
}
// Check if IP is loopback (127.0.0.1 or ::1)
if !ip.IsLoopback() {
log.Warnf("Amp management: non-localhost connection from %s attempted access, denying", remoteAddr)
c.AbortWithStatusJSON(403, gin.H{
"error": "Access denied: management routes restricted to localhost",
})
return
}
c.Next()
}
}
// noCORSMiddleware disables CORS for management routes to prevent browser-based attacks.
// This overwrites any global CORS headers set by the server.
func noCORSMiddleware() gin.HandlerFunc {
return func(c *gin.Context) {
// Remove CORS headers to prevent cross-origin access from browsers
c.Header("Access-Control-Allow-Origin", "")
c.Header("Access-Control-Allow-Methods", "")
c.Header("Access-Control-Allow-Headers", "")
c.Header("Access-Control-Allow-Credentials", "")
// For OPTIONS preflight, deny with 403
if c.Request.Method == "OPTIONS" {
c.AbortWithStatus(403)
return
}
c.Next()
}
}
// registerManagementRoutes registers Amp management proxy routes
// These routes proxy through to the Amp control plane for OAuth, user management, etc.
// If restrictToLocalhost is true, routes will only accept connections from 127.0.0.1/::1.
func (m *AmpModule) registerManagementRoutes(engine *gin.Engine, baseHandler *handlers.BaseAPIHandler, proxyHandler gin.HandlerFunc, restrictToLocalhost bool) {
ampAPI := engine.Group("/api")
// Always disable CORS for management routes to prevent browser-based attacks
ampAPI.Use(noCORSMiddleware())
// Apply localhost-only restriction if configured
if restrictToLocalhost {
ampAPI.Use(localhostOnlyMiddleware())
log.Info("Amp management routes restricted to localhost only (CORS disabled)")
} else {
log.Warn("⚠️ Amp management routes are NOT restricted to localhost - this is insecure!")
}
// Management routes - these are proxied directly to Amp upstream
ampAPI.Any("/internal", proxyHandler)
ampAPI.Any("/internal/*path", proxyHandler)
ampAPI.Any("/user", proxyHandler)
ampAPI.Any("/user/*path", proxyHandler)
ampAPI.Any("/auth", proxyHandler)
ampAPI.Any("/auth/*path", proxyHandler)
ampAPI.Any("/meta", proxyHandler)
ampAPI.Any("/meta/*path", proxyHandler)
ampAPI.Any("/ads", proxyHandler)
ampAPI.Any("/telemetry", proxyHandler)
ampAPI.Any("/telemetry/*path", proxyHandler)
ampAPI.Any("/threads", proxyHandler)
ampAPI.Any("/threads/*path", proxyHandler)
ampAPI.Any("/otel", proxyHandler)
ampAPI.Any("/otel/*path", proxyHandler)
// Google v1beta1 passthrough with OAuth fallback
// AMP CLI uses non-standard paths like /publishers/google/models/...
// We bridge these to our standard Gemini handler to enable local OAuth.
// If no local OAuth is available, falls back to ampcode.com proxy.
geminiHandlers := gemini.NewGeminiAPIHandler(baseHandler)
geminiBridge := createGeminiBridgeHandler(geminiHandlers)
geminiV1Beta1Fallback := NewFallbackHandler(func() *httputil.ReverseProxy {
return m.proxy
})
geminiV1Beta1Handler := geminiV1Beta1Fallback.WrapHandler(geminiBridge)
// Route POST model calls through Gemini bridge when a local provider exists, otherwise proxy.
// All other methods (e.g., GET model listing) always proxy to upstream to preserve Amp CLI behavior.
ampAPI.Any("/provider/google/v1beta1/*path", func(c *gin.Context) {
if c.Request.Method == "POST" {
// Attempt to extract the model name from the AMP-style path
if path := c.Param("path"); strings.Contains(path, "/models/") {
modelPart := path[strings.Index(path, "/models/")+len("/models/"):]
if colonIdx := strings.Index(modelPart, ":"); colonIdx > 0 {
modelPart = modelPart[:colonIdx]
}
if modelPart != "" {
normalized, _ := util.NormalizeGeminiThinkingModel(modelPart)
// Only handle locally when we have a provider; otherwise fall back to proxy
if providers := util.GetProviderName(normalized); len(providers) > 0 {
geminiV1Beta1Handler(c)
return
}
}
}
}
// Non-POST or no local provider available -> proxy upstream
proxyHandler(c)
})
}
// registerProviderAliases registers /api/provider/{provider}/... routes
// These allow Amp CLI to route requests like:
//
// /api/provider/openai/v1/chat/completions
// /api/provider/anthropic/v1/messages
// /api/provider/google/v1beta/models
func (m *AmpModule) registerProviderAliases(engine *gin.Engine, baseHandler *handlers.BaseAPIHandler, auth gin.HandlerFunc) {
// Create handler instances for different providers
openaiHandlers := openai.NewOpenAIAPIHandler(baseHandler)
geminiHandlers := gemini.NewGeminiAPIHandler(baseHandler)
claudeCodeHandlers := claude.NewClaudeCodeAPIHandler(baseHandler)
openaiResponsesHandlers := openai.NewOpenAIResponsesAPIHandler(baseHandler)
// Create fallback handler wrapper that forwards to ampcode.com when provider not found
// Uses lazy evaluation to access proxy (which is created after routes are registered)
fallbackHandler := NewFallbackHandler(func() *httputil.ReverseProxy {
return m.proxy
})
// Provider-specific routes under /api/provider/:provider
ampProviders := engine.Group("/api/provider")
if auth != nil {
ampProviders.Use(auth)
}
provider := ampProviders.Group("/:provider")
// Dynamic models handler - routes to appropriate provider based on path parameter
ampModelsHandler := func(c *gin.Context) {
providerName := strings.ToLower(c.Param("provider"))
switch providerName {
case "anthropic":
claudeCodeHandlers.ClaudeModels(c)
case "google":
geminiHandlers.GeminiModels(c)
default:
// Default to OpenAI-compatible (works for openai, groq, cerebras, etc.)
openaiHandlers.OpenAIModels(c)
}
}
// Root-level routes (for providers that omit /v1, like groq/cerebras)
// Wrap handlers with fallback logic to forward to ampcode.com when provider not found
provider.GET("/models", ampModelsHandler) // Models endpoint doesn't need fallback (no body to check)
provider.POST("/chat/completions", fallbackHandler.WrapHandler(openaiHandlers.ChatCompletions))
provider.POST("/completions", fallbackHandler.WrapHandler(openaiHandlers.Completions))
provider.POST("/responses", fallbackHandler.WrapHandler(openaiResponsesHandlers.Responses))
// /v1 routes (OpenAI/Claude-compatible endpoints)
v1Amp := provider.Group("/v1")
{
v1Amp.GET("/models", ampModelsHandler) // Models endpoint doesn't need fallback
// OpenAI-compatible endpoints with fallback
v1Amp.POST("/chat/completions", fallbackHandler.WrapHandler(openaiHandlers.ChatCompletions))
v1Amp.POST("/completions", fallbackHandler.WrapHandler(openaiHandlers.Completions))
v1Amp.POST("/responses", fallbackHandler.WrapHandler(openaiResponsesHandlers.Responses))
// Claude/Anthropic-compatible endpoints with fallback
v1Amp.POST("/messages", fallbackHandler.WrapHandler(claudeCodeHandlers.ClaudeMessages))
v1Amp.POST("/messages/count_tokens", fallbackHandler.WrapHandler(claudeCodeHandlers.ClaudeCountTokens))
}
// /v1beta routes (Gemini native API)
// Note: Gemini handler extracts model from URL path, so fallback logic needs special handling
v1betaAmp := provider.Group("/v1beta")
{
v1betaAmp.GET("/models", geminiHandlers.GeminiModels)
v1betaAmp.POST("/models/:action", fallbackHandler.WrapHandler(geminiHandlers.GeminiHandler))
v1betaAmp.GET("/models/:action", geminiHandlers.GeminiGetHandler)
}
}

View File

@@ -0,0 +1,301 @@
package amp
import (
"net/http"
"net/http/httptest"
"testing"
"github.com/gin-gonic/gin"
"github.com/router-for-me/CLIProxyAPI/v6/sdk/api/handlers"
)
func TestRegisterManagementRoutes(t *testing.T) {
gin.SetMode(gin.TestMode)
r := gin.New()
// Spy to track if proxy handler was called
proxyCalled := false
proxyHandler := func(c *gin.Context) {
proxyCalled = true
c.String(200, "proxied")
}
m := &AmpModule{}
base := &handlers.BaseAPIHandler{}
m.registerManagementRoutes(r, base, proxyHandler, false) // false = don't restrict to localhost in tests
managementPaths := []struct {
path string
method string
}{
{"/api/internal", http.MethodGet},
{"/api/internal/some/path", http.MethodGet},
{"/api/user", http.MethodGet},
{"/api/user/profile", http.MethodGet},
{"/api/auth", http.MethodGet},
{"/api/auth/login", http.MethodGet},
{"/api/meta", http.MethodGet},
{"/api/telemetry", http.MethodGet},
{"/api/threads", http.MethodGet},
{"/api/otel", http.MethodGet},
// Google v1beta1 bridge should still proxy non-model requests (GET) and allow POST
{"/api/provider/google/v1beta1/models", http.MethodGet},
{"/api/provider/google/v1beta1/models", http.MethodPost},
}
for _, path := range managementPaths {
t.Run(path.path, func(t *testing.T) {
proxyCalled = false
req := httptest.NewRequest(path.method, path.path, nil)
w := httptest.NewRecorder()
r.ServeHTTP(w, req)
if w.Code == http.StatusNotFound {
t.Fatalf("route %s not registered", path.path)
}
if !proxyCalled {
t.Fatalf("proxy handler not called for %s", path.path)
}
})
}
}
func TestRegisterProviderAliases_AllProvidersRegistered(t *testing.T) {
gin.SetMode(gin.TestMode)
r := gin.New()
// Minimal base handler setup (no need to initialize, just check routing)
base := &handlers.BaseAPIHandler{}
// Track if auth middleware was called
authCalled := false
authMiddleware := func(c *gin.Context) {
authCalled = true
c.Header("X-Auth", "ok")
// Abort with success to avoid calling the actual handler (which needs full setup)
c.AbortWithStatus(http.StatusOK)
}
m := &AmpModule{authMiddleware_: authMiddleware}
m.registerProviderAliases(r, base, authMiddleware)
paths := []struct {
path string
method string
}{
{"/api/provider/openai/models", http.MethodGet},
{"/api/provider/anthropic/models", http.MethodGet},
{"/api/provider/google/models", http.MethodGet},
{"/api/provider/groq/models", http.MethodGet},
{"/api/provider/openai/chat/completions", http.MethodPost},
{"/api/provider/anthropic/v1/messages", http.MethodPost},
{"/api/provider/google/v1beta/models", http.MethodGet},
}
for _, tc := range paths {
t.Run(tc.path, func(t *testing.T) {
authCalled = false
req := httptest.NewRequest(tc.method, tc.path, nil)
w := httptest.NewRecorder()
r.ServeHTTP(w, req)
if w.Code == http.StatusNotFound {
t.Fatalf("route %s %s not registered", tc.method, tc.path)
}
if !authCalled {
t.Fatalf("auth middleware not executed for %s", tc.path)
}
if w.Header().Get("X-Auth") != "ok" {
t.Fatalf("auth middleware header not set for %s", tc.path)
}
})
}
}
func TestRegisterProviderAliases_DynamicModelsHandler(t *testing.T) {
gin.SetMode(gin.TestMode)
r := gin.New()
base := &handlers.BaseAPIHandler{}
m := &AmpModule{authMiddleware_: func(c *gin.Context) { c.AbortWithStatus(http.StatusOK) }}
m.registerProviderAliases(r, base, func(c *gin.Context) { c.AbortWithStatus(http.StatusOK) })
providers := []string{"openai", "anthropic", "google", "groq", "cerebras"}
for _, provider := range providers {
t.Run(provider, func(t *testing.T) {
path := "/api/provider/" + provider + "/models"
req := httptest.NewRequest(http.MethodGet, path, nil)
w := httptest.NewRecorder()
r.ServeHTTP(w, req)
// Should not 404
if w.Code == http.StatusNotFound {
t.Fatalf("models route not found for provider: %s", provider)
}
})
}
}
func TestRegisterProviderAliases_V1Routes(t *testing.T) {
gin.SetMode(gin.TestMode)
r := gin.New()
base := &handlers.BaseAPIHandler{}
m := &AmpModule{authMiddleware_: func(c *gin.Context) { c.AbortWithStatus(http.StatusOK) }}
m.registerProviderAliases(r, base, func(c *gin.Context) { c.AbortWithStatus(http.StatusOK) })
v1Paths := []struct {
path string
method string
}{
{"/api/provider/openai/v1/models", http.MethodGet},
{"/api/provider/openai/v1/chat/completions", http.MethodPost},
{"/api/provider/openai/v1/completions", http.MethodPost},
{"/api/provider/anthropic/v1/messages", http.MethodPost},
{"/api/provider/anthropic/v1/messages/count_tokens", http.MethodPost},
}
for _, tc := range v1Paths {
t.Run(tc.path, func(t *testing.T) {
req := httptest.NewRequest(tc.method, tc.path, nil)
w := httptest.NewRecorder()
r.ServeHTTP(w, req)
if w.Code == http.StatusNotFound {
t.Fatalf("v1 route %s %s not registered", tc.method, tc.path)
}
})
}
}
func TestRegisterProviderAliases_V1BetaRoutes(t *testing.T) {
gin.SetMode(gin.TestMode)
r := gin.New()
base := &handlers.BaseAPIHandler{}
m := &AmpModule{authMiddleware_: func(c *gin.Context) { c.AbortWithStatus(http.StatusOK) }}
m.registerProviderAliases(r, base, func(c *gin.Context) { c.AbortWithStatus(http.StatusOK) })
v1betaPaths := []struct {
path string
method string
}{
{"/api/provider/google/v1beta/models", http.MethodGet},
{"/api/provider/google/v1beta/models/generateContent", http.MethodPost},
}
for _, tc := range v1betaPaths {
t.Run(tc.path, func(t *testing.T) {
req := httptest.NewRequest(tc.method, tc.path, nil)
w := httptest.NewRecorder()
r.ServeHTTP(w, req)
if w.Code == http.StatusNotFound {
t.Fatalf("v1beta route %s %s not registered", tc.method, tc.path)
}
})
}
}
func TestRegisterProviderAliases_NoAuthMiddleware(t *testing.T) {
// Test that routes still register even if auth middleware is nil (fallback behavior)
gin.SetMode(gin.TestMode)
r := gin.New()
base := &handlers.BaseAPIHandler{}
m := &AmpModule{authMiddleware_: nil} // No auth middleware
m.registerProviderAliases(r, base, func(c *gin.Context) { c.AbortWithStatus(http.StatusOK) })
req := httptest.NewRequest(http.MethodGet, "/api/provider/openai/models", nil)
w := httptest.NewRecorder()
r.ServeHTTP(w, req)
// Should still work (with fallback no-op auth)
if w.Code == http.StatusNotFound {
t.Fatal("routes should register even without auth middleware")
}
}
func TestLocalhostOnlyMiddleware_PreventsSpoofing(t *testing.T) {
gin.SetMode(gin.TestMode)
r := gin.New()
// Apply localhost-only middleware
r.Use(localhostOnlyMiddleware())
r.GET("/test", func(c *gin.Context) {
c.String(http.StatusOK, "ok")
})
tests := []struct {
name string
remoteAddr string
forwardedFor string
expectedStatus int
description string
}{
{
name: "spoofed_header_remote_connection",
remoteAddr: "192.168.1.100:12345",
forwardedFor: "127.0.0.1",
expectedStatus: http.StatusForbidden,
description: "Spoofed X-Forwarded-For header should be ignored",
},
{
name: "real_localhost_ipv4",
remoteAddr: "127.0.0.1:54321",
forwardedFor: "",
expectedStatus: http.StatusOK,
description: "Real localhost IPv4 connection should work",
},
{
name: "real_localhost_ipv6",
remoteAddr: "[::1]:54321",
forwardedFor: "",
expectedStatus: http.StatusOK,
description: "Real localhost IPv6 connection should work",
},
{
name: "remote_ipv4",
remoteAddr: "203.0.113.42:8080",
forwardedFor: "",
expectedStatus: http.StatusForbidden,
description: "Remote IPv4 connection should be blocked",
},
{
name: "remote_ipv6",
remoteAddr: "[2001:db8::1]:9090",
forwardedFor: "",
expectedStatus: http.StatusForbidden,
description: "Remote IPv6 connection should be blocked",
},
{
name: "spoofed_localhost_ipv6",
remoteAddr: "203.0.113.42:8080",
forwardedFor: "::1",
expectedStatus: http.StatusForbidden,
description: "Spoofed X-Forwarded-For with IPv6 localhost should be ignored",
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
req := httptest.NewRequest(http.MethodGet, "/test", nil)
req.RemoteAddr = tt.remoteAddr
if tt.forwardedFor != "" {
req.Header.Set("X-Forwarded-For", tt.forwardedFor)
}
w := httptest.NewRecorder()
r.ServeHTTP(w, req)
if w.Code != tt.expectedStatus {
t.Errorf("%s: expected status %d, got %d", tt.description, tt.expectedStatus, w.Code)
}
})
}
}

View File

@@ -0,0 +1,155 @@
package amp
import (
"context"
"encoding/json"
"fmt"
"os"
"path/filepath"
"strings"
"sync"
"time"
)
// SecretSource provides Amp API keys with configurable precedence and caching
type SecretSource interface {
Get(ctx context.Context) (string, error)
}
// cachedSecret holds a secret value with expiration
type cachedSecret struct {
value string
expiresAt time.Time
}
// MultiSourceSecret implements precedence-based secret lookup:
// 1. Explicit config value (highest priority)
// 2. Environment variable AMP_API_KEY
// 3. File-based secret (lowest priority)
type MultiSourceSecret struct {
explicitKey string
envKey string
filePath string
cacheTTL time.Duration
mu sync.RWMutex
cache *cachedSecret
}
// NewMultiSourceSecret creates a secret source with precedence and caching
func NewMultiSourceSecret(explicitKey string, cacheTTL time.Duration) *MultiSourceSecret {
if cacheTTL == 0 {
cacheTTL = 5 * time.Minute // Default 5 minute cache
}
home, _ := os.UserHomeDir()
filePath := filepath.Join(home, ".local", "share", "amp", "secrets.json")
return &MultiSourceSecret{
explicitKey: strings.TrimSpace(explicitKey),
envKey: "AMP_API_KEY",
filePath: filePath,
cacheTTL: cacheTTL,
}
}
// NewMultiSourceSecretWithPath creates a secret source with a custom file path (for testing)
func NewMultiSourceSecretWithPath(explicitKey string, filePath string, cacheTTL time.Duration) *MultiSourceSecret {
if cacheTTL == 0 {
cacheTTL = 5 * time.Minute
}
return &MultiSourceSecret{
explicitKey: strings.TrimSpace(explicitKey),
envKey: "AMP_API_KEY",
filePath: filePath,
cacheTTL: cacheTTL,
}
}
// Get retrieves the Amp API key using precedence: config > env > file
// Results are cached for cacheTTL duration to avoid excessive file reads
func (s *MultiSourceSecret) Get(ctx context.Context) (string, error) {
// Precedence 1: Explicit config key (highest priority, no caching needed)
if s.explicitKey != "" {
return s.explicitKey, nil
}
// Precedence 2: Environment variable
if envValue := strings.TrimSpace(os.Getenv(s.envKey)); envValue != "" {
return envValue, nil
}
// Precedence 3: File-based secret (lowest priority, cached)
// Check cache first
s.mu.RLock()
if s.cache != nil && time.Now().Before(s.cache.expiresAt) {
value := s.cache.value
s.mu.RUnlock()
return value, nil
}
s.mu.RUnlock()
// Cache miss or expired - read from file
key, err := s.readFromFile()
if err != nil {
// Cache empty result to avoid repeated file reads on missing files
s.updateCache("")
return "", err
}
// Cache the result
s.updateCache(key)
return key, nil
}
// readFromFile reads the Amp API key from the secrets file
func (s *MultiSourceSecret) readFromFile() (string, error) {
content, err := os.ReadFile(s.filePath)
if err != nil {
if os.IsNotExist(err) {
return "", nil // Missing file is not an error, just no key available
}
return "", fmt.Errorf("failed to read amp secrets from %s: %w", s.filePath, err)
}
var secrets map[string]string
if err := json.Unmarshal(content, &secrets); err != nil {
return "", fmt.Errorf("failed to parse amp secrets from %s: %w", s.filePath, err)
}
key := strings.TrimSpace(secrets["apiKey@https://ampcode.com/"])
return key, nil
}
// updateCache updates the cached secret value
func (s *MultiSourceSecret) updateCache(value string) {
s.mu.Lock()
defer s.mu.Unlock()
s.cache = &cachedSecret{
value: value,
expiresAt: time.Now().Add(s.cacheTTL),
}
}
// InvalidateCache clears the cached secret, forcing a fresh read on next Get
func (s *MultiSourceSecret) InvalidateCache() {
s.mu.Lock()
defer s.mu.Unlock()
s.cache = nil
}
// StaticSecretSource returns a fixed API key (for testing)
type StaticSecretSource struct {
key string
}
// NewStaticSecretSource creates a secret source with a fixed key
func NewStaticSecretSource(key string) *StaticSecretSource {
return &StaticSecretSource{key: strings.TrimSpace(key)}
}
// Get returns the static API key
func (s *StaticSecretSource) Get(ctx context.Context) (string, error) {
return s.key, nil
}

View File

@@ -0,0 +1,280 @@
package amp
import (
"context"
"encoding/json"
"os"
"path/filepath"
"sync"
"testing"
"time"
)
func TestMultiSourceSecret_PrecedenceOrder(t *testing.T) {
ctx := context.Background()
cases := []struct {
name string
configKey string
envKey string
fileJSON string
want string
}{
{"config_wins", "cfg", "env", `{"apiKey@https://ampcode.com/":"file"}`, "cfg"},
{"env_wins_when_no_cfg", "", "env", `{"apiKey@https://ampcode.com/":"file"}`, "env"},
{"file_when_no_cfg_env", "", "", `{"apiKey@https://ampcode.com/":"file"}`, "file"},
{"empty_cfg_trims_then_env", " ", "env", `{"apiKey@https://ampcode.com/":"file"}`, "env"},
{"empty_env_then_file", "", " ", `{"apiKey@https://ampcode.com/":"file"}`, "file"},
{"missing_file_returns_empty", "", "", "", ""},
{"all_empty_returns_empty", " ", " ", `{"apiKey@https://ampcode.com/":" "}`, ""},
}
for _, tc := range cases {
tc := tc // capture range variable
t.Run(tc.name, func(t *testing.T) {
tmpDir := t.TempDir()
secretsPath := filepath.Join(tmpDir, "secrets.json")
if tc.fileJSON != "" {
if err := os.WriteFile(secretsPath, []byte(tc.fileJSON), 0600); err != nil {
t.Fatal(err)
}
}
t.Setenv("AMP_API_KEY", tc.envKey)
s := NewMultiSourceSecretWithPath(tc.configKey, secretsPath, 100*time.Millisecond)
got, err := s.Get(ctx)
if err != nil && tc.fileJSON != "" && json.Valid([]byte(tc.fileJSON)) {
t.Fatalf("unexpected error: %v", err)
}
if got != tc.want {
t.Fatalf("want %q, got %q", tc.want, got)
}
})
}
}
func TestMultiSourceSecret_CacheBehavior(t *testing.T) {
ctx := context.Background()
tmpDir := t.TempDir()
p := filepath.Join(tmpDir, "secrets.json")
// Initial value
if err := os.WriteFile(p, []byte(`{"apiKey@https://ampcode.com/":"v1"}`), 0600); err != nil {
t.Fatal(err)
}
s := NewMultiSourceSecretWithPath("", p, 50*time.Millisecond)
// First read - should return v1
got1, err := s.Get(ctx)
if err != nil {
t.Fatalf("Get failed: %v", err)
}
if got1 != "v1" {
t.Fatalf("expected v1, got %s", got1)
}
// Change file; within TTL we should still see v1 (cached)
if err := os.WriteFile(p, []byte(`{"apiKey@https://ampcode.com/":"v2"}`), 0600); err != nil {
t.Fatal(err)
}
got2, _ := s.Get(ctx)
if got2 != "v1" {
t.Fatalf("cache hit expected v1, got %s", got2)
}
// After TTL expires, should see v2
time.Sleep(60 * time.Millisecond)
got3, _ := s.Get(ctx)
if got3 != "v2" {
t.Fatalf("cache miss expected v2, got %s", got3)
}
// Invalidate forces re-read immediately
if err := os.WriteFile(p, []byte(`{"apiKey@https://ampcode.com/":"v3"}`), 0600); err != nil {
t.Fatal(err)
}
s.InvalidateCache()
got4, _ := s.Get(ctx)
if got4 != "v3" {
t.Fatalf("invalidate expected v3, got %s", got4)
}
}
func TestMultiSourceSecret_FileHandling(t *testing.T) {
ctx := context.Background()
t.Run("missing_file_no_error", func(t *testing.T) {
s := NewMultiSourceSecretWithPath("", "/nonexistent/path/secrets.json", 100*time.Millisecond)
got, err := s.Get(ctx)
if err != nil {
t.Fatalf("expected no error for missing file, got: %v", err)
}
if got != "" {
t.Fatalf("expected empty string, got %q", got)
}
})
t.Run("invalid_json", func(t *testing.T) {
tmpDir := t.TempDir()
p := filepath.Join(tmpDir, "secrets.json")
if err := os.WriteFile(p, []byte(`{invalid json`), 0600); err != nil {
t.Fatal(err)
}
s := NewMultiSourceSecretWithPath("", p, 100*time.Millisecond)
_, err := s.Get(ctx)
if err == nil {
t.Fatal("expected error for invalid JSON")
}
})
t.Run("missing_key_in_json", func(t *testing.T) {
tmpDir := t.TempDir()
p := filepath.Join(tmpDir, "secrets.json")
if err := os.WriteFile(p, []byte(`{"other":"value"}`), 0600); err != nil {
t.Fatal(err)
}
s := NewMultiSourceSecretWithPath("", p, 100*time.Millisecond)
got, err := s.Get(ctx)
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
if got != "" {
t.Fatalf("expected empty string for missing key, got %q", got)
}
})
t.Run("empty_key_value", func(t *testing.T) {
tmpDir := t.TempDir()
p := filepath.Join(tmpDir, "secrets.json")
if err := os.WriteFile(p, []byte(`{"apiKey@https://ampcode.com/":" "}`), 0600); err != nil {
t.Fatal(err)
}
s := NewMultiSourceSecretWithPath("", p, 100*time.Millisecond)
got, _ := s.Get(ctx)
if got != "" {
t.Fatalf("expected empty after trim, got %q", got)
}
})
}
func TestMultiSourceSecret_Concurrency(t *testing.T) {
tmpDir := t.TempDir()
p := filepath.Join(tmpDir, "secrets.json")
if err := os.WriteFile(p, []byte(`{"apiKey@https://ampcode.com/":"concurrent"}`), 0600); err != nil {
t.Fatal(err)
}
s := NewMultiSourceSecretWithPath("", p, 5*time.Second)
ctx := context.Background()
// Spawn many goroutines calling Get concurrently
const goroutines = 50
const iterations = 100
var wg sync.WaitGroup
errors := make(chan error, goroutines)
for i := 0; i < goroutines; i++ {
wg.Add(1)
go func() {
defer wg.Done()
for j := 0; j < iterations; j++ {
val, err := s.Get(ctx)
if err != nil {
errors <- err
return
}
if val != "concurrent" {
errors <- err
return
}
}
}()
}
wg.Wait()
close(errors)
for err := range errors {
t.Errorf("concurrency error: %v", err)
}
}
func TestStaticSecretSource(t *testing.T) {
ctx := context.Background()
t.Run("returns_provided_key", func(t *testing.T) {
s := NewStaticSecretSource("test-key-123")
got, err := s.Get(ctx)
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
if got != "test-key-123" {
t.Fatalf("want test-key-123, got %q", got)
}
})
t.Run("trims_whitespace", func(t *testing.T) {
s := NewStaticSecretSource(" test-key ")
got, err := s.Get(ctx)
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
if got != "test-key" {
t.Fatalf("want test-key, got %q", got)
}
})
t.Run("empty_string", func(t *testing.T) {
s := NewStaticSecretSource("")
got, err := s.Get(ctx)
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
if got != "" {
t.Fatalf("want empty string, got %q", got)
}
})
}
func TestMultiSourceSecret_CacheEmptyResult(t *testing.T) {
// Test that missing file results are cached to avoid repeated file reads
tmpDir := t.TempDir()
p := filepath.Join(tmpDir, "nonexistent.json")
s := NewMultiSourceSecretWithPath("", p, 100*time.Millisecond)
ctx := context.Background()
// First call - file doesn't exist, should cache empty result
got1, err := s.Get(ctx)
if err != nil {
t.Fatalf("expected no error for missing file, got: %v", err)
}
if got1 != "" {
t.Fatalf("expected empty string, got %q", got1)
}
// Create the file now
if err := os.WriteFile(p, []byte(`{"apiKey@https://ampcode.com/":"new-value"}`), 0600); err != nil {
t.Fatal(err)
}
// Second call - should still return empty (cached), not read the new file
got2, _ := s.Get(ctx)
if got2 != "" {
t.Fatalf("cache should return empty, got %q", got2)
}
// After TTL expires, should see the new value
time.Sleep(110 * time.Millisecond)
got3, _ := s.Get(ctx)
if got3 != "new-value" {
t.Fatalf("after cache expiry, expected new-value, got %q", got3)
}
}

View File

@@ -0,0 +1,92 @@
// Package modules provides a pluggable routing module system for extending
// the API server with optional features without modifying core routing logic.
package modules
import (
"fmt"
"github.com/gin-gonic/gin"
"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
"github.com/router-for-me/CLIProxyAPI/v6/sdk/api/handlers"
)
// Context encapsulates the dependencies exposed to routing modules during
// registration. Modules can use the Gin engine to attach routes, the shared
// BaseAPIHandler for constructing SDK-specific handlers, and the resolved
// authentication middleware for protecting routes that require API keys.
type Context struct {
Engine *gin.Engine
BaseHandler *handlers.BaseAPIHandler
Config *config.Config
AuthMiddleware gin.HandlerFunc
}
// RouteModule represents a pluggable routing module that can register routes
// and handle configuration updates independently of the core server.
//
// DEPRECATED: Use RouteModuleV2 for new modules. This interface is kept for
// backwards compatibility and will be removed in a future version.
type RouteModule interface {
// Name returns a human-readable identifier for the module
Name() string
// Register sets up routes and handlers for this module.
// It receives the Gin engine, base handlers, and current configuration.
// Returns an error if registration fails (errors are logged but don't stop the server).
Register(engine *gin.Engine, baseHandler *handlers.BaseAPIHandler, cfg *config.Config) error
// OnConfigUpdated is called when the configuration is reloaded.
// Modules can respond to configuration changes here.
// Returns an error if the update cannot be applied.
OnConfigUpdated(cfg *config.Config) error
}
// RouteModuleV2 represents a pluggable bundle of routes that can integrate with
// the API server without modifying its core routing logic. Implementations can
// attach routes during Register and react to configuration updates via
// OnConfigUpdated.
//
// This is the preferred interface for new modules. It uses Context for cleaner
// dependency injection and supports idempotent registration.
type RouteModuleV2 interface {
// Name returns a unique identifier for logging and diagnostics.
Name() string
// Register wires the module's routes into the provided Gin engine. Modules
// should treat multiple calls as idempotent and avoid duplicate route
// registration when invoked more than once.
Register(ctx Context) error
// OnConfigUpdated notifies the module when the server configuration changes
// via hot reload. Implementations can refresh cached state or emit warnings.
OnConfigUpdated(cfg *config.Config) error
}
// RegisterModule is a helper that registers a module using either the V1 or V2
// interface. This allows gradual migration from V1 to V2 without breaking
// existing modules.
//
// Example usage:
//
// ctx := modules.Context{
// Engine: engine,
// BaseHandler: baseHandler,
// Config: cfg,
// AuthMiddleware: authMiddleware,
// }
// if err := modules.RegisterModule(ctx, ampModule); err != nil {
// log.Errorf("Failed to register module: %v", err)
// }
func RegisterModule(ctx Context, mod interface{}) error {
// Try V2 interface first (preferred)
if v2, ok := mod.(RouteModuleV2); ok {
return v2.Register(ctx)
}
// Fall back to V1 interface for backwards compatibility
if v1, ok := mod.(RouteModule); ok {
return v1.Register(ctx.Engine, ctx.BaseHandler, ctx.Config)
}
return fmt.Errorf("unsupported module type %T (must implement RouteModule or RouteModuleV2)", mod)
}

View File

@@ -21,6 +21,8 @@ import (
"github.com/router-for-me/CLIProxyAPI/v6/internal/access"
managementHandlers "github.com/router-for-me/CLIProxyAPI/v6/internal/api/handlers/management"
"github.com/router-for-me/CLIProxyAPI/v6/internal/api/middleware"
"github.com/router-for-me/CLIProxyAPI/v6/internal/api/modules"
ampmodule "github.com/router-for-me/CLIProxyAPI/v6/internal/api/modules/amp"
"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
"github.com/router-for-me/CLIProxyAPI/v6/internal/logging"
"github.com/router-for-me/CLIProxyAPI/v6/internal/managementasset"
@@ -261,6 +263,20 @@ func NewServer(cfg *config.Config, authManager *auth.Manager, accessManager *sdk
// Setup routes
s.setupRoutes()
// Register Amp module using V2 interface with Context
ampModule := ampmodule.NewLegacy(accessManager, AuthMiddleware(accessManager))
ctx := modules.Context{
Engine: engine,
BaseHandler: s.handlers,
Config: cfg,
AuthMiddleware: AuthMiddleware(accessManager),
}
if err := modules.RegisterModule(ctx, ampModule); err != nil {
log.Errorf("Failed to register Amp module: %v", err)
}
// Apply additional router configurators from options
if optionState.routerConfigurator != nil {
optionState.routerConfigurator(engine, s.handlers, cfg)
}
@@ -518,6 +534,7 @@ func (s *Server) registerManagementRoutes() {
mgmt.GET("/gemini-cli-auth-url", s.mgmt.RequestGeminiCLIToken)
mgmt.GET("/qwen-auth-url", s.mgmt.RequestQwenToken)
mgmt.GET("/iflow-auth-url", s.mgmt.RequestIFlowToken)
mgmt.POST("/iflow-auth-url", s.mgmt.RequestIFlowCookieToken)
mgmt.GET("/get-auth-status", s.mgmt.GetAuthStatus)
}
}

111
internal/api/server_test.go Normal file
View File

@@ -0,0 +1,111 @@
package api
import (
"net/http"
"net/http/httptest"
"os"
"path/filepath"
"strings"
"testing"
gin "github.com/gin-gonic/gin"
proxyconfig "github.com/router-for-me/CLIProxyAPI/v6/internal/config"
sdkaccess "github.com/router-for-me/CLIProxyAPI/v6/sdk/access"
"github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/auth"
sdkconfig "github.com/router-for-me/CLIProxyAPI/v6/sdk/config"
)
func newTestServer(t *testing.T) *Server {
t.Helper()
gin.SetMode(gin.TestMode)
tmpDir := t.TempDir()
authDir := filepath.Join(tmpDir, "auth")
if err := os.MkdirAll(authDir, 0o700); err != nil {
t.Fatalf("failed to create auth dir: %v", err)
}
cfg := &proxyconfig.Config{
SDKConfig: sdkconfig.SDKConfig{
APIKeys: []string{"test-key"},
},
Port: 0,
AuthDir: authDir,
Debug: true,
LoggingToFile: false,
UsageStatisticsEnabled: false,
}
authManager := auth.NewManager(nil, nil, nil)
accessManager := sdkaccess.NewManager()
configPath := filepath.Join(tmpDir, "config.yaml")
return NewServer(cfg, authManager, accessManager, configPath)
}
func TestAmpProviderModelRoutes(t *testing.T) {
testCases := []struct {
name string
path string
wantStatus int
wantContains string
}{
{
name: "openai root models",
path: "/api/provider/openai/models",
wantStatus: http.StatusOK,
wantContains: `"object":"list"`,
},
{
name: "groq root models",
path: "/api/provider/groq/models",
wantStatus: http.StatusOK,
wantContains: `"object":"list"`,
},
{
name: "openai models",
path: "/api/provider/openai/v1/models",
wantStatus: http.StatusOK,
wantContains: `"object":"list"`,
},
{
name: "anthropic models",
path: "/api/provider/anthropic/v1/models",
wantStatus: http.StatusOK,
wantContains: `"data"`,
},
{
name: "google models v1",
path: "/api/provider/google/v1/models",
wantStatus: http.StatusOK,
wantContains: `"models"`,
},
{
name: "google models v1beta",
path: "/api/provider/google/v1beta/models",
wantStatus: http.StatusOK,
wantContains: `"models"`,
},
}
for _, tc := range testCases {
tc := tc
t.Run(tc.name, func(t *testing.T) {
server := newTestServer(t)
req := httptest.NewRequest(http.MethodGet, tc.path, nil)
req.Header.Set("Authorization", "Bearer test-key")
rr := httptest.NewRecorder()
server.engine.ServeHTTP(rr, req)
if rr.Code != tc.wantStatus {
t.Fatalf("unexpected status code for %s: got %d want %d; body=%s", tc.path, rr.Code, tc.wantStatus, rr.Body.String())
}
if body := rr.Body.String(); !strings.Contains(body, tc.wantContains) {
t.Fatalf("response body for %s missing %q: %s", tc.path, tc.wantContains, body)
}
})
}
}

View File

@@ -0,0 +1,38 @@
package iflow
import (
"fmt"
"strings"
)
// NormalizeCookie normalizes raw cookie strings for iFlow authentication flows.
func NormalizeCookie(raw string) (string, error) {
trimmed := strings.TrimSpace(raw)
if trimmed == "" {
return "", fmt.Errorf("cookie cannot be empty")
}
combined := strings.Join(strings.Fields(trimmed), " ")
if !strings.HasSuffix(combined, ";") {
combined += ";"
}
if !strings.Contains(combined, "BXAuth=") {
return "", fmt.Errorf("cookie missing BXAuth field")
}
return combined, nil
}
// SanitizeIFlowFileName normalizes user identifiers for safe filename usage.
func SanitizeIFlowFileName(raw string) string {
if raw == "" {
return ""
}
cleanEmail := strings.ReplaceAll(raw, "*", "x")
var result strings.Builder
for _, r := range cleanEmail {
if (r >= 'a' && r <= 'z') || (r >= 'A' && r <= 'Z') || (r >= '0' && r <= '9') || r == '_' || r == '@' || r == '.' || r == '-' {
result.WriteRune(r)
}
}
return strings.TrimSpace(result.String())
}

View File

@@ -1,6 +1,7 @@
package iflow
import (
"compress/gzip"
"context"
"encoding/base64"
"encoding/json"
@@ -23,6 +24,9 @@ const (
iFlowUserInfoEndpoint = "https://iflow.cn/api/oauth/getUserInfo"
iFlowSuccessRedirectURL = "https://iflow.cn/oauth/success"
// Cookie authentication endpoints
iFlowAPIKeyEndpoint = "https://platform.iflow.cn/api/openapi/apikey"
// Client credentials provided by iFlow for the Code Assist integration.
iFlowOAuthClientID = "10009311001"
iFlowOAuthClientSecret = "4Z3YjXycVsQvyGF1etiNlIBB4RsqSDtW"
@@ -261,6 +265,7 @@ type IFlowTokenData struct {
Expire string
APIKey string
Email string
Cookie string
}
// userInfoResponse represents the structure returned by the user info endpoint.
@@ -274,3 +279,232 @@ type userInfoData struct {
Email string `json:"email"`
Phone string `json:"phone"`
}
// iFlowAPIKeyResponse represents the response from the API key endpoint
type iFlowAPIKeyResponse struct {
Success bool `json:"success"`
Code string `json:"code"`
Message string `json:"message"`
Data iFlowKeyData `json:"data"`
Extra interface{} `json:"extra"`
}
// iFlowKeyData contains the API key information
type iFlowKeyData struct {
HasExpired bool `json:"hasExpired"`
ExpireTime string `json:"expireTime"`
Name string `json:"name"`
APIKey string `json:"apiKey"`
APIKeyMask string `json:"apiKeyMask"`
}
// iFlowRefreshRequest represents the request body for refreshing API key
type iFlowRefreshRequest struct {
Name string `json:"name"`
}
// AuthenticateWithCookie performs authentication using browser cookies
func (ia *IFlowAuth) AuthenticateWithCookie(ctx context.Context, cookie string) (*IFlowTokenData, error) {
if strings.TrimSpace(cookie) == "" {
return nil, fmt.Errorf("iflow cookie authentication: cookie is empty")
}
// First, get initial API key information using GET request
keyInfo, err := ia.fetchAPIKeyInfo(ctx, cookie)
if err != nil {
return nil, fmt.Errorf("iflow cookie authentication: fetch initial API key info failed: %w", err)
}
// Convert to token data format
data := &IFlowTokenData{
APIKey: keyInfo.APIKey,
Expire: keyInfo.ExpireTime,
Email: keyInfo.Name,
Cookie: cookie,
}
return data, nil
}
// fetchAPIKeyInfo retrieves API key information using GET request with cookie
func (ia *IFlowAuth) fetchAPIKeyInfo(ctx context.Context, cookie string) (*iFlowKeyData, error) {
req, err := http.NewRequestWithContext(ctx, http.MethodGet, iFlowAPIKeyEndpoint, nil)
if err != nil {
return nil, fmt.Errorf("iflow cookie: create GET request failed: %w", err)
}
// Set cookie and other headers to mimic browser
req.Header.Set("Cookie", cookie)
req.Header.Set("Accept", "application/json, text/plain, */*")
req.Header.Set("User-Agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36")
req.Header.Set("Accept-Language", "zh-CN,zh;q=0.9,en;q=0.8")
req.Header.Set("Accept-Encoding", "gzip, deflate, br")
req.Header.Set("Connection", "keep-alive")
req.Header.Set("Sec-Fetch-Dest", "empty")
req.Header.Set("Sec-Fetch-Mode", "cors")
req.Header.Set("Sec-Fetch-Site", "same-origin")
resp, err := ia.httpClient.Do(req)
if err != nil {
return nil, fmt.Errorf("iflow cookie: GET request failed: %w", err)
}
defer func() { _ = resp.Body.Close() }()
// Handle gzip compression
var reader io.Reader = resp.Body
if resp.Header.Get("Content-Encoding") == "gzip" {
gzipReader, err := gzip.NewReader(resp.Body)
if err != nil {
return nil, fmt.Errorf("iflow cookie: create gzip reader failed: %w", err)
}
defer func() { _ = gzipReader.Close() }()
reader = gzipReader
}
body, err := io.ReadAll(reader)
if err != nil {
return nil, fmt.Errorf("iflow cookie: read GET response failed: %w", err)
}
if resp.StatusCode != http.StatusOK {
log.Debugf("iflow cookie GET request failed: status=%d body=%s", resp.StatusCode, string(body))
return nil, fmt.Errorf("iflow cookie: GET request failed with status %d: %s", resp.StatusCode, strings.TrimSpace(string(body)))
}
var keyResp iFlowAPIKeyResponse
if err = json.Unmarshal(body, &keyResp); err != nil {
return nil, fmt.Errorf("iflow cookie: decode GET response failed: %w", err)
}
if !keyResp.Success {
return nil, fmt.Errorf("iflow cookie: GET request not successful: %s", keyResp.Message)
}
// Handle initial response where apiKey field might be apiKeyMask
if keyResp.Data.APIKey == "" && keyResp.Data.APIKeyMask != "" {
keyResp.Data.APIKey = keyResp.Data.APIKeyMask
}
return &keyResp.Data, nil
}
// RefreshAPIKey refreshes the API key using POST request
func (ia *IFlowAuth) RefreshAPIKey(ctx context.Context, cookie, name string) (*iFlowKeyData, error) {
if strings.TrimSpace(cookie) == "" {
return nil, fmt.Errorf("iflow cookie refresh: cookie is empty")
}
if strings.TrimSpace(name) == "" {
return nil, fmt.Errorf("iflow cookie refresh: name is empty")
}
// Prepare request body
refreshReq := iFlowRefreshRequest{
Name: name,
}
bodyBytes, err := json.Marshal(refreshReq)
if err != nil {
return nil, fmt.Errorf("iflow cookie refresh: marshal request failed: %w", err)
}
req, err := http.NewRequestWithContext(ctx, http.MethodPost, iFlowAPIKeyEndpoint, strings.NewReader(string(bodyBytes)))
if err != nil {
return nil, fmt.Errorf("iflow cookie refresh: create POST request failed: %w", err)
}
// Set cookie and other headers to mimic browser
req.Header.Set("Cookie", cookie)
req.Header.Set("Content-Type", "application/json")
req.Header.Set("Accept", "application/json, text/plain, */*")
req.Header.Set("User-Agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36")
req.Header.Set("Accept-Language", "zh-CN,zh;q=0.9,en;q=0.8")
req.Header.Set("Accept-Encoding", "gzip, deflate, br")
req.Header.Set("Connection", "keep-alive")
req.Header.Set("Origin", "https://platform.iflow.cn")
req.Header.Set("Referer", "https://platform.iflow.cn/")
resp, err := ia.httpClient.Do(req)
if err != nil {
return nil, fmt.Errorf("iflow cookie refresh: POST request failed: %w", err)
}
defer func() { _ = resp.Body.Close() }()
// Handle gzip compression
var reader io.Reader = resp.Body
if resp.Header.Get("Content-Encoding") == "gzip" {
gzipReader, err := gzip.NewReader(resp.Body)
if err != nil {
return nil, fmt.Errorf("iflow cookie refresh: create gzip reader failed: %w", err)
}
defer func() { _ = gzipReader.Close() }()
reader = gzipReader
}
body, err := io.ReadAll(reader)
if err != nil {
return nil, fmt.Errorf("iflow cookie refresh: read POST response failed: %w", err)
}
if resp.StatusCode != http.StatusOK {
log.Debugf("iflow cookie POST request failed: status=%d body=%s", resp.StatusCode, string(body))
return nil, fmt.Errorf("iflow cookie refresh: POST request failed with status %d: %s", resp.StatusCode, strings.TrimSpace(string(body)))
}
var keyResp iFlowAPIKeyResponse
if err = json.Unmarshal(body, &keyResp); err != nil {
return nil, fmt.Errorf("iflow cookie refresh: decode POST response failed: %w", err)
}
if !keyResp.Success {
return nil, fmt.Errorf("iflow cookie refresh: POST request not successful: %s", keyResp.Message)
}
return &keyResp.Data, nil
}
// ShouldRefreshAPIKey checks if the API key needs to be refreshed (within 2 days of expiry)
func ShouldRefreshAPIKey(expireTime string) (bool, time.Duration, error) {
if strings.TrimSpace(expireTime) == "" {
return false, 0, fmt.Errorf("iflow cookie: expire time is empty")
}
expire, err := time.Parse("2006-01-02 15:04", expireTime)
if err != nil {
return false, 0, fmt.Errorf("iflow cookie: parse expire time failed: %w", err)
}
now := time.Now()
twoDaysFromNow := now.Add(48 * time.Hour)
needsRefresh := expire.Before(twoDaysFromNow)
timeUntilExpiry := expire.Sub(now)
return needsRefresh, timeUntilExpiry, nil
}
// CreateCookieTokenStorage converts cookie-based token data into persistence storage
func (ia *IFlowAuth) CreateCookieTokenStorage(data *IFlowTokenData) *IFlowTokenStorage {
if data == nil {
return nil
}
return &IFlowTokenStorage{
APIKey: data.APIKey,
Email: data.Email,
Expire: data.Expire,
Cookie: data.Cookie,
LastRefresh: time.Now().Format(time.RFC3339),
Type: "iflow",
}
}
// UpdateCookieTokenStorage updates the persisted token storage with refreshed API key data
func (ia *IFlowAuth) UpdateCookieTokenStorage(storage *IFlowTokenStorage, keyData *iFlowKeyData) {
if storage == nil || keyData == nil {
return
}
storage.APIKey = keyData.APIKey
storage.Expire = keyData.ExpireTime
storage.LastRefresh = time.Now().Format(time.RFC3339)
}

View File

@@ -19,6 +19,7 @@ type IFlowTokenStorage struct {
Email string `json:"email"`
TokenType string `json:"token_type"`
Scope string `json:"scope"`
Cookie string `json:"cookie"`
Type string `json:"type"`
}

View File

@@ -0,0 +1,38 @@
package cmd
import (
"context"
"fmt"
"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
sdkAuth "github.com/router-for-me/CLIProxyAPI/v6/sdk/auth"
log "github.com/sirupsen/logrus"
)
// DoAntigravityLogin triggers the OAuth flow for the antigravity provider and saves tokens.
func DoAntigravityLogin(cfg *config.Config, options *LoginOptions) {
if options == nil {
options = &LoginOptions{}
}
manager := newAuthManager()
authOpts := &sdkAuth.LoginOptions{
NoBrowser: options.NoBrowser,
Metadata: map[string]string{},
Prompt: options.Prompt,
}
record, savedPath, err := manager.Login(context.Background(), "antigravity", cfg, authOpts)
if err != nil {
log.Errorf("Antigravity authentication failed: %v", err)
return
}
if savedPath != "" {
fmt.Printf("Authentication saved to %s\n", savedPath)
}
if record != nil && record.Label != "" {
fmt.Printf("Authenticated as %s\n", record.Label)
}
fmt.Println("Antigravity authentication successful!")
}

View File

@@ -18,6 +18,7 @@ func newAuthManager() *sdkAuth.Manager {
sdkAuth.NewClaudeAuthenticator(),
sdkAuth.NewQwenAuthenticator(),
sdkAuth.NewIFlowAuthenticator(),
sdkAuth.NewAntigravityAuthenticator(),
)
return manager
}

View File

@@ -0,0 +1,86 @@
package cmd
import (
"bufio"
"context"
"fmt"
"os"
"strings"
"github.com/router-for-me/CLIProxyAPI/v6/internal/auth/iflow"
"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
)
// DoIFlowCookieAuth performs the iFlow cookie-based authentication.
func DoIFlowCookieAuth(cfg *config.Config, options *LoginOptions) {
if options == nil {
options = &LoginOptions{}
}
promptFn := options.Prompt
if promptFn == nil {
reader := bufio.NewReader(os.Stdin)
promptFn = func(prompt string) (string, error) {
fmt.Print(prompt)
value, err := reader.ReadString('\n')
if err != nil {
return "", err
}
return strings.TrimSpace(value), nil
}
}
// Prompt user for cookie
cookie, err := promptForCookie(promptFn)
if err != nil {
fmt.Printf("Failed to get cookie: %v\n", err)
return
}
// Authenticate with cookie
auth := iflow.NewIFlowAuth(cfg)
ctx := context.Background()
tokenData, err := auth.AuthenticateWithCookie(ctx, cookie)
if err != nil {
fmt.Printf("iFlow cookie authentication failed: %v\n", err)
return
}
// Create token storage
tokenStorage := auth.CreateCookieTokenStorage(tokenData)
// Get auth file path using email in filename
authFilePath := getAuthFilePath(cfg, "iflow", tokenData.Email)
// Save token to file
if err := tokenStorage.SaveTokenToFile(authFilePath); err != nil {
fmt.Printf("Failed to save authentication: %v\n", err)
return
}
fmt.Printf("Authentication successful! API key: %s\n", tokenData.APIKey)
fmt.Printf("Expires at: %s\n", tokenData.Expire)
fmt.Printf("Authentication saved to: %s\n", authFilePath)
}
// promptForCookie prompts the user to enter their iFlow cookie
func promptForCookie(promptFn func(string) (string, error)) (string, error) {
line, err := promptFn("Enter iFlow Cookie (from browser cookies): ")
if err != nil {
return "", fmt.Errorf("failed to read cookie: %w", err)
}
cookie, err := iflow.NormalizeCookie(line)
if err != nil {
return "", err
}
return cookie, nil
}
// getAuthFilePath returns the auth file path for the given provider and email
func getAuthFilePath(cfg *config.Config, provider, email string) string {
fileName := iflow.SanitizeIFlowFileName(email)
return fmt.Sprintf("%s/%s-%s.json", cfg.AuthDir, provider, fileName)
}

View File

@@ -23,6 +23,17 @@ type Config struct {
// Port is the network port on which the API server will listen.
Port int `yaml:"port" json:"-"`
// AmpUpstreamURL defines the upstream Amp control plane used for non-provider calls.
AmpUpstreamURL string `yaml:"amp-upstream-url" json:"amp-upstream-url"`
// AmpUpstreamAPIKey optionally overrides the Authorization header when proxying Amp upstream calls.
AmpUpstreamAPIKey string `yaml:"amp-upstream-api-key" json:"amp-upstream-api-key"`
// AmpRestrictManagementToLocalhost restricts Amp management routes (/api/user, /api/threads, etc.)
// to only accept connections from localhost (127.0.0.1, ::1). When true, prevents drive-by
// browser attacks and remote access to management endpoints. Default: true (recommended).
AmpRestrictManagementToLocalhost bool `yaml:"amp-restrict-management-to-localhost" json:"amp-restrict-management-to-localhost"`
// AuthDir is the directory where authentication token files are stored.
AuthDir string `yaml:"auth-dir" json:"-"`
@@ -258,6 +269,7 @@ func LoadConfigOptional(configFile string, optional bool) (*Config, error) {
cfg.LoggingToFile = false
cfg.UsageStatisticsEnabled = false
cfg.DisableCooling = false
cfg.AmpRestrictManagementToLocalhost = true // Default to secure: only localhost access
if err = yaml.Unmarshal(data, &cfg); err != nil {
if optional {
// In cloud deploy mode, if YAML parsing fails, return empty config instead of error.

View File

@@ -62,6 +62,9 @@ type Part struct {
// InlineData contains base64-encoded data with its MIME type (e.g., images).
InlineData *InlineData `json:"inlineData,omitempty"`
// ThoughtSignature is a provider-required signature that accompanies certain parts.
ThoughtSignature string `json:"thoughtSignature,omitempty"`
// FunctionCall represents a tool call requested by the model.
FunctionCall *FunctionCall `json:"functionCall,omitempty"`

View File

@@ -17,6 +17,7 @@ func CodexInstructionsForModel(modelName, systemInstructions string) (bool, stri
lastPrompt := ""
lastCodexPrompt := ""
lastCodexMaxPrompt := ""
last51Prompt := ""
// lastReviewPrompt := ""
for _, entry := range entries {
@@ -26,6 +27,8 @@ func CodexInstructionsForModel(modelName, systemInstructions string) (bool, stri
}
if strings.HasPrefix(entry.Name(), "gpt_5_codex_prompt.md") {
lastCodexPrompt = string(content)
} else if strings.HasPrefix(entry.Name(), "gpt-5.1-codex-max_prompt.md") {
lastCodexMaxPrompt = string(content)
} else if strings.HasPrefix(entry.Name(), "prompt.md") {
lastPrompt = string(content)
} else if strings.HasPrefix(entry.Name(), "gpt_5_1_prompt.md") {
@@ -34,8 +37,9 @@ func CodexInstructionsForModel(modelName, systemInstructions string) (bool, stri
// lastReviewPrompt = string(content)
}
}
if strings.Contains(modelName, "codex") {
if strings.Contains(modelName, "codex-max") {
return false, lastCodexMaxPrompt
} else if strings.Contains(modelName, "codex") {
return false, lastCodexPrompt
} else if strings.Contains(modelName, "5.1") {
return false, last51Prompt

View File

@@ -0,0 +1,117 @@
You are Codex, based on GPT-5. You are running as a coding agent in the Codex CLI on a user's computer.
## General
- When searching for text or files, prefer using `rg` or `rg --files` respectively because `rg` is much faster than alternatives like `grep`. (If the `rg` command is not found, then use alternatives.)
## Editing constraints
- Default to ASCII when editing or creating files. Only introduce non-ASCII or other Unicode characters when there is a clear justification and the file already uses them.
- Add succinct code comments that explain what is going on if code is not self-explanatory. You should not add comments like "Assigns the value to the variable", but a brief comment might be useful ahead of a complex code block that the user would otherwise have to spend time parsing out. Usage of these comments should be rare.
- Try to use apply_patch for single file edits, but it is fine to explore other options to make the edit if it does not work well. Do not use apply_patch for changes that are auto-generated (i.e. generating package.json or running a lint or format command like gofmt) or when scripting is more efficient (such as search and replacing a string across a codebase).
- You may be in a dirty git worktree.
* NEVER revert existing changes you did not make unless explicitly requested, since these changes were made by the user.
* If asked to make a commit or code edits and there are unrelated changes to your work or changes that you didn't make in those files, don't revert those changes.
* If the changes are in files you've touched recently, you should read carefully and understand how you can work with the changes rather than reverting them.
* If the changes are in unrelated files, just ignore them and don't revert them.
- Do not amend a commit unless explicitly requested to do so.
- While you are working, you might notice unexpected changes that you didn't make. If this happens, STOP IMMEDIATELY and ask the user how they would like to proceed.
- **NEVER** use destructive commands like `git reset --hard` or `git checkout --` unless specifically requested or approved by the user.
## Plan tool
When using the planning tool:
- Skip using the planning tool for straightforward tasks (roughly the easiest 25%).
- Do not make single-step plans.
- When you made a plan, update it after having performed one of the sub-tasks that you shared on the plan.
## Codex CLI harness, sandboxing, and approvals
The Codex CLI harness supports several different configurations for sandboxing and escalation approvals that the user can choose from.
Filesystem sandboxing defines which files can be read or written. The options for `sandbox_mode` are:
- **read-only**: The sandbox only permits reading files.
- **workspace-write**: The sandbox permits reading files, and editing files in `cwd` and `writable_roots`. Editing files in other directories requires approval.
- **danger-full-access**: No filesystem sandboxing - all commands are permitted.
Network sandboxing defines whether network can be accessed without approval. Options for `network_access` are:
- **restricted**: Requires approval
- **enabled**: No approval needed
Approvals are your mechanism to get user consent to run shell commands without the sandbox. Possible configuration options for `approval_policy` are
- **untrusted**: The harness will escalate most commands for user approval, apart from a limited allowlist of safe "read" commands.
- **on-failure**: The harness will allow all commands to run in the sandbox (if enabled), and failures will be escalated to the user for approval to run again without the sandbox.
- **on-request**: Commands will be run in the sandbox by default, and you can specify in your tool call if you want to escalate a command to run without sandboxing. (Note that this mode is not always available. If it is, you'll see parameters for it in the `shell` command description.)
- **never**: This is a non-interactive mode where you may NEVER ask the user for approval to run commands. Instead, you must always persist and work around constraints to solve the task for the user. You MUST do your utmost best to finish the task and validate your work before yielding. If this mode is paired with `danger-full-access`, take advantage of it to deliver the best outcome for the user. Further, in this mode, your default testing philosophy is overridden: Even if you don't see local patterns for testing, you may add tests and scripts to validate your work. Just remove them before yielding.
When you are running with `approval_policy == on-request`, and sandboxing enabled, here are scenarios where you'll need to request approval:
- You need to run a command that writes to a directory that requires it (e.g. running tests that write to /var)
- You need to run a GUI app (e.g., open/xdg-open/osascript) to open browsers or files.
- You are running sandboxed and need to run a command that requires network access (e.g. installing packages)
- If you run a command that is important to solving the user's query, but it fails because of sandboxing, rerun the command with approval. ALWAYS proceed to use the `with_escalated_permissions` and `justification` parameters - do not message the user before requesting approval for the command.
- You are about to take a potentially destructive action such as an `rm` or `git reset` that the user did not explicitly ask for
- (for all of these, you should weigh alternative paths that do not require approval)
When `sandbox_mode` is set to read-only, you'll need to request approval for any command that isn't a read.
You will be told what filesystem sandboxing, network sandboxing, and approval mode are active in a developer or user message. If you are not told about this, assume that you are running with workspace-write, network sandboxing enabled, and approval on-failure.
Although they introduce friction to the user because your work is paused until the user responds, you should leverage them when necessary to accomplish important work. If the completing the task requires escalated permissions, Do not let these settings or the sandbox deter you from attempting to accomplish the user's task unless it is set to "never", in which case never ask for approvals.
When requesting approval to execute a command that will require escalated privileges:
- Provide the `with_escalated_permissions` parameter with the boolean value true
- Include a short, 1 sentence explanation for why you need to enable `with_escalated_permissions` in the justification parameter
## Special user requests
- If the user makes a simple request (such as asking for the time) which you can fulfill by running a terminal command (such as `date`), you should do so.
- If the user asks for a "review", default to a code review mindset: prioritise identifying bugs, risks, behavioural regressions, and missing tests. Findings must be the primary focus of the response - keep summaries or overviews brief and only after enumerating the issues. Present findings first (ordered by severity with file/line references), follow with open questions or assumptions, and offer a change-summary only as a secondary detail. If no findings are discovered, state that explicitly and mention any residual risks or testing gaps.
## Frontend tasks
When doing frontend design tasks, avoid collapsing into "AI slop" or safe, average-looking layouts.
Aim for interfaces that feel intentional, bold, and a bit surprising.
- Typography: Use expressive, purposeful fonts and avoid default stacks (Inter, Roboto, Arial, system).
- Color & Look: Choose a clear visual direction; define CSS variables; avoid purple-on-white defaults. No purple bias or dark mode bias.
- Motion: Use a few meaningful animations (page-load, staggered reveals) instead of generic micro-motions.
- Background: Don't rely on flat, single-color backgrounds; use gradients, shapes, or subtle patterns to build atmosphere.
- Overall: Avoid boilerplate layouts and interchangeable UI patterns. Vary themes, type families, and visual languages across outputs.
- Ensure the page loads properly on both desktop and mobile
Exception: If working within an existing website or design system, preserve the established patterns, structure, and visual language.
## Presenting your work and final message
You are producing plain text that will later be styled by the CLI. Follow these rules exactly. Formatting should make results easy to scan, but not feel mechanical. Use judgment to decide how much structure adds value.
- Default: be very concise; friendly coding teammate tone.
- Ask only when needed; suggest ideas; mirror the user's style.
- For substantial work, summarize clearly; follow finalanswer formatting.
- Skip heavy formatting for simple confirmations.
- Don't dump large files you've written; reference paths only.
- No "save/copy this file" - User is on the same machine.
- Offer logical next steps (tests, commits, build) briefly; add verify steps if you couldn't do something.
- For code changes:
* Lead with a quick explanation of the change, and then give more details on the context covering where and why a change was made. Do not start this explanation with "summary", just jump right in.
* If there are natural next steps the user may want to take, suggest them at the end of your response. Do not make suggestions if there are no natural next steps.
* When suggesting multiple options, use numeric lists for the suggestions so the user can quickly respond with a single number.
- The user does not command execution outputs. When asked to show the output of a command (e.g. `git show`), relay the important details in your answer or summarize the key lines so the user understands the result.
### Final answer structure and style guidelines
- Plain text; CLI handles styling. Use structure only when it helps scanability.
- Headers: optional; short Title Case (1-3 words) wrapped in **…**; no blank line before the first bullet; add only if they truly help.
- Bullets: use - ; merge related points; keep to one line when possible; 46 per list ordered by importance; keep phrasing consistent.
- Monospace: backticks for commands/paths/env vars/code ids and inline examples; use for literal keyword bullets; never combine with **.
- Code samples or multi-line snippets should be wrapped in fenced code blocks; include an info string as often as possible.
- Structure: group related bullets; order sections general → specific → supporting; for subsections, start with a bolded keyword bullet, then items; match complexity to the task.
- Tone: collaborative, concise, factual; present tense, active voice; selfcontained; no "above/below"; parallel wording.
- Don'ts: no nested bullets/hierarchies; no ANSI codes; don't cram unrelated keywords; keep keyword lists short—wrap/reformat if long; avoid naming formatting styles in answers.
- Adaptation: code explanations → precise, structured with code refs; simple tasks → lead with outcome; big changes → logical walkthrough + rationale + next actions; casual one-offs → plain sentences, no headers/bullets.
- File References: When referencing files in your response follow the below rules:
* Use inline code to make file paths clickable.
* Each reference should have a stand alone path. Even if it's the same file.
* Accepted: absolute, workspacerelative, a/ or b/ diff prefixes, or bare filename/suffix.
* Optionally include line/column (1based): :line[:column] or #Lline[Ccolumn] (column defaults to 1).
* Do not use URIs like file://, vscode://, or https://.
* Do not provide range of lines
* Examples: src/app.ts, src/app.ts:42, b/server/index.js#L10, C:\repo\project\main.rs:12:5

View File

@@ -3,8 +3,6 @@
// when registering their supported models.
package registry
import "time"
// GetClaudeModels returns the standard Claude model definitions
func GetClaudeModels() []*ModelInfo {
return []*ModelInfo{
@@ -68,28 +66,13 @@ func GetClaudeModels() []*ModelInfo {
}
}
// GeminiModels returns the shared base Gemini model set used by multiple providers.
func GeminiModels() []*ModelInfo {
// GetGeminiModels returns the standard Gemini model definitions
func GetGeminiModels() []*ModelInfo {
return []*ModelInfo{
{
ID: "gemini-2.5-flash",
Object: "model",
Created: time.Now().Unix(),
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-flash",
Version: "001",
DisplayName: "Gemini 2.5 Flash",
Description: "Stable version of Gemini 2.5 Flash, our mid-size multimodal model that supports up to 1 million tokens, released in June of 2025.",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-2.5-pro",
Object: "model",
Created: time.Now().Unix(),
Created: 1750118400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-pro",
@@ -101,10 +84,25 @@ func GeminiModels() []*ModelInfo {
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 128, Max: 32768, ZeroAllowed: false, DynamicAllowed: true},
},
{
ID: "gemini-2.5-flash",
Object: "model",
Created: 1750118400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-flash",
Version: "001",
DisplayName: "Gemini 2.5 Flash",
Description: "Stable version of Gemini 2.5 Flash, our mid-size multimodal model that supports up to 1 million tokens, released in June of 2025.",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-2.5-flash-lite",
Object: "model",
Created: time.Now().Unix(),
Created: 1753142400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-flash-lite",
@@ -114,36 +112,32 @@ func GeminiModels() []*ModelInfo {
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 512, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-3-pro-preview",
Object: "model",
Created: 1737158400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-3-pro-preview",
Version: "3.0",
DisplayName: "Gemini 3 Pro Preview",
Description: "Gemini 3 Pro Preview",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 128, Max: 32768, ZeroAllowed: false, DynamicAllowed: true},
},
}
}
// GetGeminiModels returns the standard Gemini model definitions
func GetGeminiModels() []*ModelInfo { return GeminiModels() }
// GetGeminiCLIModels returns the standard Gemini model definitions
func GetGeminiCLIModels() []*ModelInfo {
func GetGeminiVertexModels() []*ModelInfo {
return []*ModelInfo{
{
ID: "gemini-2.5-flash",
Object: "model",
Created: time.Now().Unix(),
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-flash",
Version: "001",
DisplayName: "Gemini 2.5 Flash",
Description: "Stable version of Gemini 2.5 Flash, our mid-size multimodal model that supports up to 1 million tokens, released in June of 2025.",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-2.5-pro",
Object: "model",
Created: time.Now().Unix(),
Created: 1750118400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-pro",
@@ -156,15 +150,125 @@ func GetGeminiCLIModels() []*ModelInfo {
Thinking: &ThinkingSupport{Min: 128, Max: 32768, ZeroAllowed: false, DynamicAllowed: true},
},
{
ID: "gemini-3-pro-preview-11-2025",
ID: "gemini-2.5-flash",
Object: "model",
Created: time.Now().Unix(),
Created: 1750118400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-3-pro-preview-11-2025",
Version: "3",
DisplayName: "Gemini 3 Pro Preview 11-2025",
Description: "Latest preview of Gemini Pro",
Name: "models/gemini-2.5-flash",
Version: "001",
DisplayName: "Gemini 2.5 Flash",
Description: "Stable version of Gemini 2.5 Flash, our mid-size multimodal model that supports up to 1 million tokens, released in June of 2025.",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-2.5-flash-lite",
Object: "model",
Created: 1753142400,
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"},
Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-3-pro-preview",
Object: "model",
Created: 1737158400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-3-pro-preview",
Version: "3.0",
DisplayName: "Gemini 3 Pro Preview",
Description: "Gemini 3 Pro Preview",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 128, Max: 32768, ZeroAllowed: false, DynamicAllowed: true},
},
{
ID: "gemini-3-pro-image-preview",
Object: "model",
Created: 1737158400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-3-pro-image-preview",
Version: "3.0",
DisplayName: "Gemini 3 Pro Image Preview",
Description: "Gemini 3 Pro Image Preview",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
},
}
}
// GetGeminiCLIModels returns the standard Gemini model definitions
func GetGeminiCLIModels() []*ModelInfo {
return []*ModelInfo{
{
ID: "gemini-2.5-pro",
Object: "model",
Created: 1750118400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-pro",
Version: "2.5",
DisplayName: "Gemini 2.5 Pro",
Description: "Stable release (June 17th, 2025) of Gemini 2.5 Pro",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 128, Max: 32768, ZeroAllowed: false, DynamicAllowed: true},
},
{
ID: "gemini-2.5-flash",
Object: "model",
Created: 1750118400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-flash",
Version: "001",
DisplayName: "Gemini 2.5 Flash",
Description: "Stable version of Gemini 2.5 Flash, our mid-size multimodal model that supports up to 1 million tokens, released in June of 2025.",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-2.5-flash-lite",
Object: "model",
Created: 1753142400,
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"},
Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-3-pro-preview",
Object: "model",
Created: 1737158400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-3-pro-preview",
Version: "3.0",
DisplayName: "Gemini 3 Pro Preview",
Description: "Gemini 3 Pro Preview",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
@@ -175,87 +279,143 @@ func GetGeminiCLIModels() []*ModelInfo {
// GetAIStudioModels returns the Gemini model definitions for AI Studio integrations
func GetAIStudioModels() []*ModelInfo {
base := GeminiModels()
return append(base,
[]*ModelInfo{
{
ID: "gemini-pro-latest",
Object: "model",
Created: time.Now().Unix(),
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-pro-latest",
Version: "2.5",
DisplayName: "Gemini Pro Latest",
Description: "Latest release of Gemini Pro",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 128, Max: 32768, ZeroAllowed: false, DynamicAllowed: true},
},
{
ID: "gemini-flash-latest",
Object: "model",
Created: time.Now().Unix(),
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-flash-latest",
Version: "2.5",
DisplayName: "Gemini Flash Latest",
Description: "Latest release of Gemini Flash",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-flash-lite-latest",
Object: "model",
Created: time.Now().Unix(),
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-flash-lite-latest",
Version: "2.5",
DisplayName: "Gemini Flash-Lite Latest",
Description: "Latest release of Gemini Flash-Lite",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 512, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-2.5-flash-image-preview",
Object: "model",
Created: time.Now().Unix(),
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-flash-image-preview",
Version: "2.5",
DisplayName: "Gemini 2.5 Flash Image Preview",
Description: "State-of-the-art image generation and editing model.",
InputTokenLimit: 1048576,
OutputTokenLimit: 8192,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
// image models don't support thinkingConfig; leave Thinking nil
},
{
ID: "gemini-2.5-flash-image",
Object: "model",
Created: time.Now().Unix(),
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-flash-image",
Version: "2.5",
DisplayName: "Gemini 2.5 Flash Image",
Description: "State-of-the-art image generation and editing model.",
InputTokenLimit: 1048576,
OutputTokenLimit: 8192,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
// image models don't support thinkingConfig; leave Thinking nil
},
}...,
)
return []*ModelInfo{
{
ID: "gemini-2.5-pro",
Object: "model",
Created: 1750118400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-pro",
Version: "2.5",
DisplayName: "Gemini 2.5 Pro",
Description: "Stable release (June 17th, 2025) of Gemini 2.5 Pro",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 128, Max: 32768, ZeroAllowed: false, DynamicAllowed: true},
},
{
ID: "gemini-2.5-flash",
Object: "model",
Created: 1750118400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-flash",
Version: "001",
DisplayName: "Gemini 2.5 Flash",
Description: "Stable version of Gemini 2.5 Flash, our mid-size multimodal model that supports up to 1 million tokens, released in June of 2025.",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-2.5-flash-lite",
Object: "model",
Created: 1753142400,
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"},
Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-3-pro-preview",
Object: "model",
Created: 1737158400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-3-pro-preview",
Version: "3.0",
DisplayName: "Gemini 3 Pro Preview",
Description: "Gemini 3 Pro Preview",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 128, Max: 32768, ZeroAllowed: false, DynamicAllowed: true},
},
{
ID: "gemini-pro-latest",
Object: "model",
Created: 1750118400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-pro-latest",
Version: "2.5",
DisplayName: "Gemini Pro Latest",
Description: "Latest release of Gemini Pro",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 128, Max: 32768, ZeroAllowed: false, DynamicAllowed: true},
},
{
ID: "gemini-flash-latest",
Object: "model",
Created: 1750118400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-flash-latest",
Version: "2.5",
DisplayName: "Gemini Flash Latest",
Description: "Latest release of Gemini Flash",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-flash-lite-latest",
Object: "model",
Created: 1753142400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-flash-lite-latest",
Version: "2.5",
DisplayName: "Gemini Flash-Lite Latest",
Description: "Latest release of Gemini Flash-Lite",
InputTokenLimit: 1048576,
OutputTokenLimit: 65536,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
Thinking: &ThinkingSupport{Min: 512, Max: 24576, ZeroAllowed: true, DynamicAllowed: true},
},
{
ID: "gemini-2.5-flash-image-preview",
Object: "model",
Created: 1756166400,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-flash-image-preview",
Version: "2.5",
DisplayName: "Gemini 2.5 Flash Image Preview",
Description: "State-of-the-art image generation and editing model.",
InputTokenLimit: 1048576,
OutputTokenLimit: 8192,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
// image models don't support thinkingConfig; leave Thinking nil
},
{
ID: "gemini-2.5-flash-image",
Object: "model",
Created: 1759363200,
OwnedBy: "google",
Type: "gemini",
Name: "models/gemini-2.5-flash-image",
Version: "2.5",
DisplayName: "Gemini 2.5 Flash Image",
Description: "State-of-the-art image generation and editing model.",
InputTokenLimit: 1048576,
OutputTokenLimit: 8192,
SupportedGenerationMethods: []string{"generateContent", "countTokens", "createCachedContent", "batchGenerateContent"},
// image models don't support thinkingConfig; leave Thinking nil
},
}
}
// GetOpenAIModels returns the standard OpenAI model definitions
@@ -264,7 +424,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5",
Object: "model",
Created: time.Now().Unix(),
Created: 1754524800,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5-2025-08-07",
@@ -277,7 +437,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5-minimal",
Object: "model",
Created: time.Now().Unix(),
Created: 1754524800,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5-2025-08-07",
@@ -290,7 +450,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5-low",
Object: "model",
Created: time.Now().Unix(),
Created: 1754524800,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5-2025-08-07",
@@ -303,7 +463,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5-medium",
Object: "model",
Created: time.Now().Unix(),
Created: 1754524800,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5-2025-08-07",
@@ -316,7 +476,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5-high",
Object: "model",
Created: time.Now().Unix(),
Created: 1754524800,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5-2025-08-07",
@@ -329,7 +489,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5-codex",
Object: "model",
Created: time.Now().Unix(),
Created: 1757894400,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5-2025-09-15",
@@ -342,7 +502,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5-codex-low",
Object: "model",
Created: time.Now().Unix(),
Created: 1757894400,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5-2025-09-15",
@@ -355,7 +515,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5-codex-medium",
Object: "model",
Created: time.Now().Unix(),
Created: 1757894400,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5-2025-09-15",
@@ -368,7 +528,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5-codex-high",
Object: "model",
Created: time.Now().Unix(),
Created: 1757894400,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5-2025-09-15",
@@ -381,7 +541,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5-codex-mini",
Object: "model",
Created: time.Now().Unix(),
Created: 1762473600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5-2025-11-07",
@@ -394,7 +554,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5-codex-mini-medium",
Object: "model",
Created: time.Now().Unix(),
Created: 1762473600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5-2025-11-07",
@@ -407,7 +567,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5-codex-mini-high",
Object: "model",
Created: time.Now().Unix(),
Created: 1762473600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5-2025-11-07",
@@ -420,7 +580,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5.1",
Object: "model",
Created: time.Now().Unix(),
Created: 1762905600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-2025-11-12",
@@ -433,7 +593,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5.1-none",
Object: "model",
Created: time.Now().Unix(),
Created: 1762905600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-2025-11-12",
@@ -446,7 +606,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5.1-low",
Object: "model",
Created: time.Now().Unix(),
Created: 1762905600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-2025-11-12",
@@ -459,7 +619,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5.1-medium",
Object: "model",
Created: time.Now().Unix(),
Created: 1762905600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-2025-11-12",
@@ -472,7 +632,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5.1-high",
Object: "model",
Created: time.Now().Unix(),
Created: 1762905600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-2025-11-12",
@@ -485,7 +645,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5.1-codex",
Object: "model",
Created: time.Now().Unix(),
Created: 1762905600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-2025-11-12",
@@ -498,7 +658,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5.1-codex-low",
Object: "model",
Created: time.Now().Unix(),
Created: 1762905600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-2025-11-12",
@@ -511,7 +671,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5.1-codex-medium",
Object: "model",
Created: time.Now().Unix(),
Created: 1762905600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-2025-11-12",
@@ -524,7 +684,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5.1-codex-high",
Object: "model",
Created: time.Now().Unix(),
Created: 1762905600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-2025-11-12",
@@ -537,7 +697,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5.1-codex-mini",
Object: "model",
Created: time.Now().Unix(),
Created: 1762905600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-2025-11-12",
@@ -550,7 +710,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5.1-codex-mini-medium",
Object: "model",
Created: time.Now().Unix(),
Created: 1762905600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-2025-11-12",
@@ -563,7 +723,7 @@ func GetOpenAIModels() []*ModelInfo {
{
ID: "gpt-5.1-codex-mini-high",
Object: "model",
Created: time.Now().Unix(),
Created: 1762905600,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-2025-11-12",
@@ -573,6 +733,72 @@ func GetOpenAIModels() []*ModelInfo {
MaxCompletionTokens: 128000,
SupportedParameters: []string{"tools"},
},
{
ID: "gpt-5.1-codex-max",
Object: "model",
Created: 1763424000,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-max",
DisplayName: "GPT 5 Codex Max",
Description: "Stable version of GPT 5 Codex Max",
ContextLength: 400000,
MaxCompletionTokens: 128000,
SupportedParameters: []string{"tools"},
},
{
ID: "gpt-5.1-codex-max-low",
Object: "model",
Created: 1763424000,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-max",
DisplayName: "GPT 5 Codex Max Low",
Description: "Stable version of GPT 5 Codex Max Low",
ContextLength: 400000,
MaxCompletionTokens: 128000,
SupportedParameters: []string{"tools"},
},
{
ID: "gpt-5.1-codex-max-medium",
Object: "model",
Created: 1763424000,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-max",
DisplayName: "GPT 5 Codex Max Medium",
Description: "Stable version of GPT 5 Codex Max Medium",
ContextLength: 400000,
MaxCompletionTokens: 128000,
SupportedParameters: []string{"tools"},
},
{
ID: "gpt-5.1-codex-max-high",
Object: "model",
Created: 1763424000,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-max",
DisplayName: "GPT 5 Codex Max High",
Description: "Stable version of GPT 5 Codex Max High",
ContextLength: 400000,
MaxCompletionTokens: 128000,
SupportedParameters: []string{"tools"},
},
{
ID: "gpt-5.1-codex-max-xhigh",
Object: "model",
Created: 1763424000,
OwnedBy: "openai",
Type: "openai",
Version: "gpt-5.1-max",
DisplayName: "GPT 5 Codex Max XHigh",
Description: "Stable version of GPT 5 Codex Max XHigh",
ContextLength: 400000,
MaxCompletionTokens: 128000,
SupportedParameters: []string{"tools"},
},
}
}
@@ -582,7 +808,7 @@ func GetQwenModels() []*ModelInfo {
{
ID: "qwen3-coder-plus",
Object: "model",
Created: time.Now().Unix(),
Created: 1753228800,
OwnedBy: "qwen",
Type: "qwen",
Version: "3.0",
@@ -595,7 +821,7 @@ func GetQwenModels() []*ModelInfo {
{
ID: "qwen3-coder-flash",
Object: "model",
Created: time.Now().Unix(),
Created: 1753228800,
OwnedBy: "qwen",
Type: "qwen",
Version: "3.0",
@@ -608,7 +834,7 @@ func GetQwenModels() []*ModelInfo {
{
ID: "vision-model",
Object: "model",
Created: time.Now().Unix(),
Created: 1758672000,
OwnedBy: "qwen",
Type: "qwen",
Version: "3.0",
@@ -624,38 +850,38 @@ func GetQwenModels() []*ModelInfo {
// GetIFlowModels returns supported models for iFlow OAuth accounts.
func GetIFlowModels() []*ModelInfo {
created := time.Now().Unix()
entries := []struct {
ID string
DisplayName string
Description string
Created int64
}{
{ID: "tstars2.0", DisplayName: "TStars-2.0", Description: "iFlow TStars-2.0 multimodal assistant"},
{ID: "qwen3-coder-plus", DisplayName: "Qwen3-Coder-Plus", Description: "Qwen3 Coder Plus code generation"},
{ID: "qwen3-coder", DisplayName: "Qwen3-Coder-480B-A35B", Description: "Qwen3 Coder 480B A35B"},
{ID: "qwen3-max", DisplayName: "Qwen3-Max", Description: "Qwen3 flagship model"},
{ID: "qwen3-vl-plus", DisplayName: "Qwen3-VL-Plus", Description: "Qwen3 multimodal vision-language"},
{ID: "qwen3-max-preview", DisplayName: "Qwen3-Max-Preview", Description: "Qwen3 Max preview build"},
{ID: "kimi-k2-0905", DisplayName: "Kimi-K2-Instruct-0905", Description: "Moonshot Kimi K2 instruct 0905"},
{ID: "glm-4.6", DisplayName: "GLM-4.6", Description: "Zhipu GLM 4.6 general model"},
{ID: "kimi-k2", DisplayName: "Kimi-K2", Description: "Moonshot Kimi K2 general model"},
{ID: "kimi-k2-thinking", DisplayName: "Kimi-K2-Thinking", Description: "Moonshot Kimi K2 general model"},
{ID: "deepseek-v3.2", DisplayName: "DeepSeek-V3.2-Exp", Description: "DeepSeek V3.2 experimental"},
{ID: "deepseek-v3.1", DisplayName: "DeepSeek-V3.1-Terminus", Description: "DeepSeek V3.1 Terminus"},
{ID: "deepseek-r1", DisplayName: "DeepSeek-R1", Description: "DeepSeek reasoning model R1"},
{ID: "deepseek-v3", DisplayName: "DeepSeek-V3-671B", Description: "DeepSeek V3 671B"},
{ID: "qwen3-32b", DisplayName: "Qwen3-32B", Description: "Qwen3 32B"},
{ID: "qwen3-235b-a22b-thinking-2507", DisplayName: "Qwen3-235B-A22B-Thinking", Description: "Qwen3 235B A22B Thinking (2507)"},
{ID: "qwen3-235b-a22b-instruct", DisplayName: "Qwen3-235B-A22B-Instruct", Description: "Qwen3 235B A22B Instruct"},
{ID: "qwen3-235b", DisplayName: "Qwen3-235B-A22B", Description: "Qwen3 235B A22B"},
{ID: "minimax-m2", DisplayName: "MiniMax-M2", Description: "MiniMax M2"},
{ID: "tstars2.0", DisplayName: "TStars-2.0", Description: "iFlow TStars-2.0 multimodal assistant", Created: 1746489600},
{ID: "qwen3-coder-plus", DisplayName: "Qwen3-Coder-Plus", Description: "Qwen3 Coder Plus code generation", Created: 1753228800},
{ID: "qwen3-coder", DisplayName: "Qwen3-Coder-480B-A35B", Description: "Qwen3 Coder 480B A35B", Created: 1753228800},
{ID: "qwen3-max", DisplayName: "Qwen3-Max", Description: "Qwen3 flagship model", Created: 1758672000},
{ID: "qwen3-vl-plus", DisplayName: "Qwen3-VL-Plus", Description: "Qwen3 multimodal vision-language", Created: 1758672000},
{ID: "qwen3-max-preview", DisplayName: "Qwen3-Max-Preview", Description: "Qwen3 Max preview build", Created: 1757030400},
{ID: "kimi-k2-0905", DisplayName: "Kimi-K2-Instruct-0905", Description: "Moonshot Kimi K2 instruct 0905", Created: 1757030400},
{ID: "glm-4.6", DisplayName: "GLM-4.6", Description: "Zhipu GLM 4.6 general model", Created: 1759190400},
{ID: "kimi-k2", DisplayName: "Kimi-K2", Description: "Moonshot Kimi K2 general model", Created: 1752192000},
{ID: "kimi-k2-thinking", DisplayName: "Kimi-K2-Thinking", Description: "Moonshot Kimi K2 general model", Created: 1762387200},
{ID: "deepseek-v3.2", DisplayName: "DeepSeek-V3.2-Exp", Description: "DeepSeek V3.2 experimental", Created: 1759104000},
{ID: "deepseek-v3.1", DisplayName: "DeepSeek-V3.1-Terminus", Description: "DeepSeek V3.1 Terminus", Created: 1756339200},
{ID: "deepseek-r1", DisplayName: "DeepSeek-R1", Description: "DeepSeek reasoning model R1", Created: 1737331200},
{ID: "deepseek-v3", DisplayName: "DeepSeek-V3-671B", Description: "DeepSeek V3 671B", Created: 1734307200},
{ID: "qwen3-32b", DisplayName: "Qwen3-32B", Description: "Qwen3 32B", Created: 1747094400},
{ID: "qwen3-235b-a22b-thinking-2507", DisplayName: "Qwen3-235B-A22B-Thinking", Description: "Qwen3 235B A22B Thinking (2507)", Created: 1753401600},
{ID: "qwen3-235b-a22b-instruct", DisplayName: "Qwen3-235B-A22B-Instruct", Description: "Qwen3 235B A22B Instruct", Created: 1753401600},
{ID: "qwen3-235b", DisplayName: "Qwen3-235B-A22B", Description: "Qwen3 235B A22B", Created: 1753401600},
{ID: "minimax-m2", DisplayName: "MiniMax-M2", Description: "MiniMax M2", Created: 1758672000},
}
models := make([]*ModelInfo, 0, len(entries))
for _, entry := range entries {
models = append(models, &ModelInfo{
ID: entry.ID,
Object: "model",
Created: created,
Created: entry.Created,
OwnedBy: "iflow",
Type: "iflow",
DisplayName: entry.DisplayName,

View File

@@ -151,7 +151,7 @@ func (e *AIStudioExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth
case wsrelay.MessageTypeStreamChunk:
if len(event.Payload) > 0 {
appendAPIResponseChunk(ctx, e.cfg, bytes.Clone(event.Payload))
filtered := filterAIStudioUsageMetadata(event.Payload)
filtered := FilterSSEUsageMetadata(event.Payload)
if detail, ok := parseGeminiStreamUsage(filtered); ok {
reporter.publish(ctx, detail)
}
@@ -264,6 +264,7 @@ func (e *AIStudioExecutor) translateRequest(req cliproxyexecutor.Request, opts c
}
payload = util.ApplyGeminiThinkingConfig(payload, budgetOverride, includeOverride)
}
payload = util.ConvertThinkingLevelToBudget(payload)
payload = util.StripThinkingConfigIfUnsupported(req.Model, payload)
payload = fixGeminiImageAspectRatio(req.Model, payload)
payload = applyPayloadConfig(e.cfg, req.Model, payload)
@@ -295,65 +296,6 @@ func (e *AIStudioExecutor) buildEndpoint(model, action, alt string) string {
return base
}
// filterAIStudioUsageMetadata removes usageMetadata from intermediate SSE events so that
// only the terminal chunk retains token statistics.
func filterAIStudioUsageMetadata(payload []byte) []byte {
if len(payload) == 0 {
return payload
}
lines := bytes.Split(payload, []byte("\n"))
modified := false
for idx, line := range lines {
trimmed := bytes.TrimSpace(line)
if len(trimmed) == 0 || !bytes.HasPrefix(trimmed, []byte("data:")) {
continue
}
dataIdx := bytes.Index(line, []byte("data:"))
if dataIdx < 0 {
continue
}
rawJSON := bytes.TrimSpace(line[dataIdx+5:])
cleaned, changed := stripUsageMetadataFromJSON(rawJSON)
if !changed {
continue
}
var rebuilt []byte
rebuilt = append(rebuilt, line[:dataIdx]...)
rebuilt = append(rebuilt, []byte("data:")...)
if len(cleaned) > 0 {
rebuilt = append(rebuilt, ' ')
rebuilt = append(rebuilt, cleaned...)
}
lines[idx] = rebuilt
modified = true
}
if !modified {
return payload
}
return bytes.Join(lines, []byte("\n"))
}
// stripUsageMetadataFromJSON drops usageMetadata when no finishReason is present.
func stripUsageMetadataFromJSON(rawJSON []byte) ([]byte, bool) {
jsonBytes := bytes.TrimSpace(rawJSON)
if len(jsonBytes) == 0 || !gjson.ValidBytes(jsonBytes) {
return rawJSON, false
}
finishReason := gjson.GetBytes(jsonBytes, "candidates.0.finishReason")
if finishReason.Exists() && finishReason.String() != "" {
return rawJSON, false
}
if !gjson.GetBytes(jsonBytes, "usageMetadata").Exists() {
return rawJSON, false
}
cleaned, err := sjson.DeleteBytes(jsonBytes, "usageMetadata")
if err != nil {
return rawJSON, false
}
return cleaned, true
}
// ensureColonSpacedJSON normalizes JSON objects so that colons are followed by a single space while
// keeping the payload otherwise compact. Non-JSON inputs are returned unchanged.
func ensureColonSpacedJSON(payload []byte) []byte {

View File

@@ -0,0 +1,565 @@
package executor
import (
"bufio"
"bytes"
"context"
"encoding/json"
"fmt"
"io"
"math/rand"
"net/http"
"net/url"
"strconv"
"strings"
"time"
"github.com/google/uuid"
"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
"github.com/router-for-me/CLIProxyAPI/v6/internal/registry"
cliproxyauth "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/auth"
cliproxyexecutor "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/executor"
sdktranslator "github.com/router-for-me/CLIProxyAPI/v6/sdk/translator"
log "github.com/sirupsen/logrus"
"github.com/tidwall/gjson"
"github.com/tidwall/sjson"
)
const (
antigravityBaseURL = "https://daily-cloudcode-pa.sandbox.googleapis.com"
antigravityStreamPath = "/v1internal:streamGenerateContent"
antigravityGeneratePath = "/v1internal:generateContent"
antigravityModelsPath = "/v1internal:fetchAvailableModels"
antigravityClientID = "1071006060591-tmhssin2h21lcre235vtolojh4g403ep.apps.googleusercontent.com"
antigravityClientSecret = "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"
defaultAntigravityAgent = "antigravity/1.11.3 windows/amd64"
antigravityAuthType = "antigravity"
refreshSkew = 5 * time.Minute
streamScannerBuffer int = 20_971_520
)
var randSource = rand.New(rand.NewSource(time.Now().UnixNano()))
// AntigravityExecutor proxies requests to the antigravity upstream.
type AntigravityExecutor struct {
cfg *config.Config
}
// NewAntigravityExecutor constructs a new executor instance.
func NewAntigravityExecutor(cfg *config.Config) *AntigravityExecutor {
return &AntigravityExecutor{cfg: cfg}
}
// Identifier implements ProviderExecutor.
func (e *AntigravityExecutor) Identifier() string { return antigravityAuthType }
// PrepareRequest implements ProviderExecutor.
func (e *AntigravityExecutor) PrepareRequest(_ *http.Request, _ *cliproxyauth.Auth) error { return nil }
// Execute handles non-streaming requests via the antigravity generate endpoint.
func (e *AntigravityExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (resp cliproxyexecutor.Response, err error) {
token, updatedAuth, errToken := e.ensureAccessToken(ctx, auth)
if errToken != nil {
return resp, errToken
}
if updatedAuth != nil {
auth = updatedAuth
}
reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
defer reporter.trackFailure(ctx, &err)
from := opts.SourceFormat
to := sdktranslator.FromString("gemini-cli")
translated := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), false)
httpReq, errReq := e.buildRequest(ctx, auth, token, req.Model, translated, false, opts.Alt)
if errReq != nil {
return resp, errReq
}
httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
httpResp, errDo := httpClient.Do(httpReq)
if errDo != nil {
recordAPIResponseError(ctx, e.cfg, errDo)
return resp, errDo
}
defer func() {
if errClose := httpResp.Body.Close(); errClose != nil {
log.Errorf("antigravity executor: close response body error: %v", errClose)
}
}()
recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
bodyBytes, errRead := io.ReadAll(httpResp.Body)
if errRead != nil {
recordAPIResponseError(ctx, e.cfg, errRead)
return resp, errRead
}
appendAPIResponseChunk(ctx, e.cfg, bodyBytes)
if httpResp.StatusCode < http.StatusOK || httpResp.StatusCode >= http.StatusMultipleChoices {
log.Debugf("antigravity executor: upstream error status: %d, body: %s", httpResp.StatusCode, summarizeErrorBody(httpResp.Header.Get("Content-Type"), bodyBytes))
err = statusErr{code: httpResp.StatusCode, msg: string(bodyBytes)}
return resp, err
}
var param any
converted := sdktranslator.TranslateNonStream(ctx, to, from, req.Model, bytes.Clone(opts.OriginalRequest), translated, bodyBytes, &param)
resp = cliproxyexecutor.Response{Payload: []byte(converted)}
reporter.ensurePublished(ctx)
return resp, nil
}
// ExecuteStream handles streaming requests via the antigravity upstream.
func (e *AntigravityExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (stream <-chan cliproxyexecutor.StreamChunk, err error) {
ctx = context.WithValue(ctx, "alt", "")
token, updatedAuth, errToken := e.ensureAccessToken(ctx, auth)
if errToken != nil {
return nil, errToken
}
if updatedAuth != nil {
auth = updatedAuth
}
reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
defer reporter.trackFailure(ctx, &err)
from := opts.SourceFormat
to := sdktranslator.FromString("gemini-cli")
translated := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), true)
httpReq, errReq := e.buildRequest(ctx, auth, token, req.Model, translated, true, opts.Alt)
if errReq != nil {
return nil, errReq
}
httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
httpResp, errDo := httpClient.Do(httpReq)
if errDo != nil {
recordAPIResponseError(ctx, e.cfg, errDo)
return nil, errDo
}
recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
if httpResp.StatusCode < http.StatusOK || httpResp.StatusCode >= http.StatusMultipleChoices {
bodyBytes, _ := io.ReadAll(httpResp.Body)
appendAPIResponseChunk(ctx, e.cfg, bodyBytes)
if errClose := httpResp.Body.Close(); errClose != nil {
log.Errorf("antigravity executor: close response body error: %v", errClose)
}
err = statusErr{code: httpResp.StatusCode, msg: string(bodyBytes)}
return nil, err
}
out := make(chan cliproxyexecutor.StreamChunk)
stream = out
go func() {
defer close(out)
defer func() {
if errClose := httpResp.Body.Close(); errClose != nil {
log.Errorf("antigravity executor: close response body error: %v", errClose)
}
}()
scanner := bufio.NewScanner(httpResp.Body)
scanner.Buffer(nil, streamScannerBuffer)
var param any
for scanner.Scan() {
line := scanner.Bytes()
appendAPIResponseChunk(ctx, e.cfg, line)
// Filter usage metadata for all models
// Only retain usage statistics in the terminal chunk
line = FilterSSEUsageMetadata(line)
chunks := sdktranslator.TranslateStream(ctx, to, from, req.Model, bytes.Clone(opts.OriginalRequest), translated, bytes.Clone(line), &param)
for i := range chunks {
out <- cliproxyexecutor.StreamChunk{Payload: []byte(chunks[i])}
}
}
tail := sdktranslator.TranslateStream(ctx, to, from, req.Model, bytes.Clone(opts.OriginalRequest), translated, []byte("[DONE]"), &param)
for i := range tail {
out <- cliproxyexecutor.StreamChunk{Payload: []byte(tail[i])}
}
if errScan := scanner.Err(); errScan != nil {
recordAPIResponseError(ctx, e.cfg, errScan)
reporter.publishFailure(ctx)
out <- cliproxyexecutor.StreamChunk{Err: errScan}
} else {
reporter.ensurePublished(ctx)
}
}()
return stream, nil
}
// Refresh refreshes the OAuth token using the refresh token.
func (e *AntigravityExecutor) Refresh(ctx context.Context, auth *cliproxyauth.Auth) (*cliproxyauth.Auth, error) {
if auth == nil {
return auth, nil
}
updated, errRefresh := e.refreshToken(ctx, auth.Clone())
if errRefresh != nil {
return nil, errRefresh
}
return updated, nil
}
// CountTokens is not supported for the antigravity provider.
func (e *AntigravityExecutor) CountTokens(context.Context, *cliproxyauth.Auth, cliproxyexecutor.Request, cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
return cliproxyexecutor.Response{}, statusErr{code: http.StatusNotImplemented, msg: "count tokens not supported"}
}
// FetchAntigravityModels retrieves available models using the supplied auth.
func FetchAntigravityModels(ctx context.Context, auth *cliproxyauth.Auth, cfg *config.Config) []*registry.ModelInfo {
exec := &AntigravityExecutor{cfg: cfg}
token, updatedAuth, errToken := exec.ensureAccessToken(ctx, auth)
if errToken != nil || token == "" {
return nil
}
if updatedAuth != nil {
auth = updatedAuth
}
modelsURL := buildBaseURL(auth) + antigravityModelsPath
httpReq, errReq := http.NewRequestWithContext(ctx, http.MethodPost, modelsURL, bytes.NewReader([]byte(`{}`)))
if errReq != nil {
return nil
}
httpReq.Header.Set("Content-Type", "application/json")
httpReq.Header.Set("Authorization", "Bearer "+token)
httpReq.Header.Set("User-Agent", resolveUserAgent(auth))
if host := resolveHost(auth); host != "" {
httpReq.Host = host
}
httpClient := newProxyAwareHTTPClient(ctx, cfg, auth, 0)
httpResp, errDo := httpClient.Do(httpReq)
if errDo != nil {
return nil
}
defer func() {
if errClose := httpResp.Body.Close(); errClose != nil {
log.Errorf("antigravity executor: close response body error: %v", errClose)
}
}()
bodyBytes, errRead := io.ReadAll(httpResp.Body)
if errRead != nil {
return nil
}
if httpResp.StatusCode < http.StatusOK || httpResp.StatusCode >= http.StatusMultipleChoices {
return nil
}
result := gjson.GetBytes(bodyBytes, "models")
if !result.Exists() {
return nil
}
now := time.Now().Unix()
models := make([]*registry.ModelInfo, 0, len(result.Map()))
for id := range result.Map() {
models = append(models, &registry.ModelInfo{
ID: id,
Object: "model",
Created: now,
OwnedBy: antigravityAuthType,
Type: antigravityAuthType,
})
}
return models
}
func (e *AntigravityExecutor) ensureAccessToken(ctx context.Context, auth *cliproxyauth.Auth) (string, *cliproxyauth.Auth, error) {
if auth == nil {
return "", nil, statusErr{code: http.StatusUnauthorized, msg: "missing auth"}
}
accessToken := metaStringValue(auth.Metadata, "access_token")
expiry := tokenExpiry(auth.Metadata)
if accessToken != "" && expiry.After(time.Now().Add(refreshSkew)) {
return accessToken, nil, nil
}
updated, errRefresh := e.refreshToken(ctx, auth.Clone())
if errRefresh != nil {
return "", nil, errRefresh
}
return metaStringValue(updated.Metadata, "access_token"), updated, nil
}
func (e *AntigravityExecutor) refreshToken(ctx context.Context, auth *cliproxyauth.Auth) (*cliproxyauth.Auth, error) {
if auth == nil {
return nil, statusErr{code: http.StatusUnauthorized, msg: "missing auth"}
}
refreshToken := metaStringValue(auth.Metadata, "refresh_token")
if refreshToken == "" {
return auth, statusErr{code: http.StatusUnauthorized, msg: "missing refresh token"}
}
form := url.Values{}
form.Set("client_id", antigravityClientID)
form.Set("client_secret", antigravityClientSecret)
form.Set("grant_type", "refresh_token")
form.Set("refresh_token", refreshToken)
httpReq, errReq := http.NewRequestWithContext(ctx, http.MethodPost, "https://oauth2.googleapis.com/token", strings.NewReader(form.Encode()))
if errReq != nil {
return auth, errReq
}
httpReq.Header.Set("Host", "oauth2.googleapis.com")
httpReq.Header.Set("User-Agent", defaultAntigravityAgent)
httpReq.Header.Set("Content-Type", "application/x-www-form-urlencoded")
httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
httpResp, errDo := httpClient.Do(httpReq)
if errDo != nil {
return auth, errDo
}
defer func() {
if errClose := httpResp.Body.Close(); errClose != nil {
log.Errorf("antigravity executor: close response body error: %v", errClose)
}
}()
bodyBytes, errRead := io.ReadAll(httpResp.Body)
if errRead != nil {
return auth, errRead
}
if httpResp.StatusCode < http.StatusOK || httpResp.StatusCode >= http.StatusMultipleChoices {
return auth, statusErr{code: httpResp.StatusCode, msg: string(bodyBytes)}
}
var tokenResp struct {
AccessToken string `json:"access_token"`
RefreshToken string `json:"refresh_token"`
ExpiresIn int64 `json:"expires_in"`
TokenType string `json:"token_type"`
}
if errUnmarshal := json.Unmarshal(bodyBytes, &tokenResp); errUnmarshal != nil {
return auth, errUnmarshal
}
if auth.Metadata == nil {
auth.Metadata = make(map[string]any)
}
auth.Metadata["access_token"] = tokenResp.AccessToken
if tokenResp.RefreshToken != "" {
auth.Metadata["refresh_token"] = tokenResp.RefreshToken
}
auth.Metadata["expires_in"] = tokenResp.ExpiresIn
auth.Metadata["timestamp"] = time.Now().UnixMilli()
auth.Metadata["expired"] = time.Now().Add(time.Duration(tokenResp.ExpiresIn) * time.Second).Format(time.RFC3339)
auth.Metadata["type"] = antigravityAuthType
return auth, nil
}
func (e *AntigravityExecutor) buildRequest(ctx context.Context, auth *cliproxyauth.Auth, token, modelName string, payload []byte, stream bool, alt string) (*http.Request, error) {
if token == "" {
return nil, statusErr{code: http.StatusUnauthorized, msg: "missing access token"}
}
base := buildBaseURL(auth)
path := antigravityGeneratePath
if stream {
path = antigravityStreamPath
}
var requestURL strings.Builder
requestURL.WriteString(base)
requestURL.WriteString(path)
if stream {
if alt != "" {
requestURL.WriteString("?$alt=")
requestURL.WriteString(url.QueryEscape(alt))
} else {
requestURL.WriteString("?alt=sse")
}
} else if alt != "" {
requestURL.WriteString("?$alt=")
requestURL.WriteString(url.QueryEscape(alt))
}
payload = geminiToAntigravity(modelName, payload)
httpReq, errReq := http.NewRequestWithContext(ctx, http.MethodPost, requestURL.String(), bytes.NewReader(payload))
if errReq != nil {
return nil, errReq
}
httpReq.Header.Set("Content-Type", "application/json")
httpReq.Header.Set("Authorization", "Bearer "+token)
httpReq.Header.Set("User-Agent", resolveUserAgent(auth))
if stream {
httpReq.Header.Set("Accept", "text/event-stream")
} else {
httpReq.Header.Set("Accept", "application/json")
}
if host := resolveHost(auth); host != "" {
httpReq.Host = host
}
var authID, authLabel, authType, authValue string
if auth != nil {
authID = auth.ID
authLabel = auth.Label
authType, authValue = auth.AccountInfo()
}
recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
URL: requestURL.String(),
Method: http.MethodPost,
Headers: httpReq.Header.Clone(),
Body: payload,
Provider: e.Identifier(),
AuthID: authID,
AuthLabel: authLabel,
AuthType: authType,
AuthValue: authValue,
})
return httpReq, nil
}
func tokenExpiry(metadata map[string]any) time.Time {
if metadata == nil {
return time.Time{}
}
if expStr, ok := metadata["expired"].(string); ok {
expStr = strings.TrimSpace(expStr)
if expStr != "" {
if parsed, errParse := time.Parse(time.RFC3339, expStr); errParse == nil {
return parsed
}
}
}
expiresIn, hasExpires := int64Value(metadata["expires_in"])
tsMs, hasTimestamp := int64Value(metadata["timestamp"])
if hasExpires && hasTimestamp {
return time.Unix(0, tsMs*int64(time.Millisecond)).Add(time.Duration(expiresIn) * time.Second)
}
return time.Time{}
}
func metaStringValue(metadata map[string]any, key string) string {
if metadata == nil {
return ""
}
if v, ok := metadata[key]; ok {
switch typed := v.(type) {
case string:
return strings.TrimSpace(typed)
case []byte:
return strings.TrimSpace(string(typed))
}
}
return ""
}
func int64Value(value any) (int64, bool) {
switch typed := value.(type) {
case int:
return int64(typed), true
case int64:
return typed, true
case float64:
return int64(typed), true
case json.Number:
if i, errParse := typed.Int64(); errParse == nil {
return i, true
}
case string:
if strings.TrimSpace(typed) == "" {
return 0, false
}
if i, errParse := strconv.ParseInt(strings.TrimSpace(typed), 10, 64); errParse == nil {
return i, true
}
}
return 0, false
}
func buildBaseURL(auth *cliproxyauth.Auth) string {
if auth != nil {
if auth.Attributes != nil {
if v := strings.TrimSpace(auth.Attributes["base_url"]); v != "" {
return strings.TrimSuffix(v, "/")
}
}
if auth.Metadata != nil {
if v, ok := auth.Metadata["base_url"].(string); ok {
v = strings.TrimSpace(v)
if v != "" {
return strings.TrimSuffix(v, "/")
}
}
}
}
return antigravityBaseURL
}
func resolveHost(auth *cliproxyauth.Auth) string {
base := buildBaseURL(auth)
parsed, errParse := url.Parse(base)
if errParse != nil {
return ""
}
if parsed.Host != "" {
return parsed.Host
}
return strings.TrimPrefix(strings.TrimPrefix(base, "https://"), "http://")
}
func resolveUserAgent(auth *cliproxyauth.Auth) string {
if auth != nil {
if auth.Attributes != nil {
if ua := strings.TrimSpace(auth.Attributes["user_agent"]); ua != "" {
return ua
}
}
if auth.Metadata != nil {
if ua, ok := auth.Metadata["user_agent"].(string); ok && strings.TrimSpace(ua) != "" {
return strings.TrimSpace(ua)
}
}
}
return defaultAntigravityAgent
}
func geminiToAntigravity(modelName string, payload []byte) []byte {
template, _ := sjson.Set(string(payload), "model", modelName)
template, _ = sjson.Set(template, "userAgent", "antigravity")
template, _ = sjson.Set(template, "project", generateProjectID())
template, _ = sjson.Set(template, "requestId", generateRequestID())
template, _ = sjson.Set(template, "request.sessionId", generateSessionID())
template, _ = sjson.Delete(template, "request.safetySettings")
template, _ = sjson.Set(template, "request.toolConfig.functionCallingConfig.mode", "VALIDATED")
gjson.Get(template, "request.contents").ForEach(func(key, content gjson.Result) bool {
if content.Get("role").String() == "model" {
content.Get("parts").ForEach(func(partKey, part gjson.Result) bool {
if part.Get("functionCall").Exists() {
template, _ = sjson.Set(template, fmt.Sprintf("request.contents.%d.parts.%d.thoughtSignature", key.Int(), partKey.Int()), "skip_thought_signature_validator")
}
return true
})
}
return true
})
return []byte(template)
}
func generateRequestID() string {
return "agent-" + uuid.NewString()
}
func generateSessionID() string {
n := randSource.Int63n(9_000_000_000_000_000_000)
return "-" + strconv.FormatInt(n, 10)
}
func generateProjectID() string {
adjectives := []string{"useful", "bright", "swift", "calm", "bold"}
nouns := []string{"fuze", "wave", "spark", "flow", "core"}
adj := adjectives[randSource.Intn(len(adjectives))]
noun := nouns[randSource.Intn(len(nouns))]
randomPart := strings.ToLower(uuid.NewString())[:5]
return adj + "-" + noun + "-" + randomPart
}

View File

@@ -219,8 +219,7 @@ func (e *ClaudeExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.A
// If from == to (Claude → Claude), directly forward the SSE stream without translation
if from == to {
scanner := bufio.NewScanner(decodedBody)
buf := make([]byte, 20_971_520)
scanner.Buffer(buf, 20_971_520)
scanner.Buffer(nil, 20_971_520)
for scanner.Scan() {
line := scanner.Bytes()
appendAPIResponseChunk(ctx, e.cfg, line)
@@ -243,8 +242,7 @@ func (e *ClaudeExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.A
// For other formats, use translation
scanner := bufio.NewScanner(decodedBody)
buf := make([]byte, 20_971_520)
scanner.Buffer(buf, 20_971_520)
scanner.Buffer(nil, 20_971_520)
var param any
for scanner.Scan() {
line := scanner.Bytes()

View File

@@ -205,8 +205,7 @@ func (e *CodexExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Au
}
}()
scanner := bufio.NewScanner(httpResp.Body)
buf := make([]byte, 20_971_520)
scanner.Buffer(buf, 20_971_520)
scanner.Buffer(nil, 20_971_520)
var param any
for scanner.Scan() {
line := scanner.Bytes()
@@ -323,6 +322,18 @@ func (e *CodexExecutor) setReasoningEffortByAlias(modelName string, payload []by
case "gpt-5.1-codex-mini-high":
payload, _ = sjson.SetBytes(payload, "reasoning.effort", "high")
}
} else if util.InArray([]string{"gpt-5.1-codex-max", "gpt-5.1-codex-max-low", "gpt-5.1-codex-max-medium", "gpt-5.1-codex-max-high", "gpt-5.1-codex-max-xhigh"}, modelName) {
payload, _ = sjson.SetBytes(payload, "model", "gpt-5.1-codex-max")
switch modelName {
case "gpt-5.1-codex-max-low":
payload, _ = sjson.SetBytes(payload, "reasoning.effort", "low")
case "gpt-5.1-codex-max-medium":
payload, _ = sjson.SetBytes(payload, "reasoning.effort", "medium")
case "gpt-5.1-codex-max-high":
payload, _ = sjson.SetBytes(payload, "reasoning.effort", "high")
case "gpt-5.1-codex-max-xhigh":
payload, _ = sjson.SetBytes(payload, "reasoning.effort", "xhigh")
}
}
return payload
}

View File

@@ -180,7 +180,7 @@ func (e *GeminiCLIExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth
continue
}
err = statusErr{code: httpResp.StatusCode, msg: string(data)}
err = newGeminiStatusErr(httpResp.StatusCode, data)
return resp, err
}
@@ -190,7 +190,7 @@ func (e *GeminiCLIExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth
if lastStatus == 0 {
lastStatus = 429
}
err = statusErr{code: lastStatus, msg: string(lastBody)}
err = newGeminiStatusErr(lastStatus, lastBody)
return resp, err
}
@@ -304,7 +304,7 @@ func (e *GeminiCLIExecutor) ExecuteStream(ctx context.Context, auth *cliproxyaut
}
continue
}
err = statusErr{code: httpResp.StatusCode, msg: string(data)}
err = newGeminiStatusErr(httpResp.StatusCode, data)
return nil, err
}
@@ -319,8 +319,7 @@ func (e *GeminiCLIExecutor) ExecuteStream(ctx context.Context, auth *cliproxyaut
}()
if opts.Alt == "" {
scanner := bufio.NewScanner(resp.Body)
buf := make([]byte, 20_971_520)
scanner.Buffer(buf, 20_971_520)
scanner.Buffer(nil, 20_971_520)
var param any
for scanner.Scan() {
line := scanner.Bytes()
@@ -378,7 +377,7 @@ func (e *GeminiCLIExecutor) ExecuteStream(ctx context.Context, auth *cliproxyaut
if lastStatus == 0 {
lastStatus = 429
}
err = statusErr{code: lastStatus, msg: string(lastBody)}
err = newGeminiStatusErr(lastStatus, lastBody)
return nil, err
}
@@ -486,7 +485,7 @@ func (e *GeminiCLIExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.
if lastStatus == 0 {
lastStatus = 429
}
return cliproxyexecutor.Response{}, statusErr{code: lastStatus, msg: string(lastBody)}
return cliproxyexecutor.Response{}, newGeminiStatusErr(lastStatus, lastBody)
}
func (e *GeminiCLIExecutor) Refresh(ctx context.Context, auth *cliproxyauth.Auth) (*cliproxyauth.Auth, error) {
@@ -770,3 +769,42 @@ func fixGeminiCLIImageAspectRatio(modelName string, rawJSON []byte) []byte {
}
return rawJSON
}
func newGeminiStatusErr(statusCode int, body []byte) statusErr {
err := statusErr{code: statusCode, msg: string(body)}
if statusCode == http.StatusTooManyRequests {
if retryAfter, parseErr := parseRetryDelay(body); parseErr == nil && retryAfter != nil {
err.retryAfter = retryAfter
}
}
return err
}
// parseRetryDelay extracts the retry delay from a Google API 429 error response.
// The error response contains a RetryInfo.retryDelay field in the format "0.847655010s".
// Returns the parsed duration or an error if it cannot be determined.
func parseRetryDelay(errorBody []byte) (*time.Duration, error) {
// Try to parse the retryDelay from the error response
// Format: error.details[].retryDelay where @type == "type.googleapis.com/google.rpc.RetryInfo"
details := gjson.GetBytes(errorBody, "error.details")
if !details.Exists() || !details.IsArray() {
return nil, fmt.Errorf("no error.details found")
}
for _, detail := range details.Array() {
typeVal := detail.Get("@type").String()
if typeVal == "type.googleapis.com/google.rpc.RetryInfo" {
retryDelay := detail.Get("retryDelay").String()
if retryDelay != "" {
// Parse duration string like "0.847655010s"
duration, err := time.ParseDuration(retryDelay)
if err != nil {
return nil, fmt.Errorf("failed to parse duration")
}
return &duration, nil
}
}
}
return nil, fmt.Errorf("no RetryInfo found")
}

View File

@@ -251,8 +251,7 @@ func (e *GeminiExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.A
}
}()
scanner := bufio.NewScanner(httpResp.Body)
buf := make([]byte, 20_971_520)
scanner.Buffer(buf, 20_971_520)
scanner.Buffer(nil, 20_971_520)
var param any
for scanner.Scan() {
line := scanner.Bytes()

View File

@@ -240,8 +240,7 @@ func (e *GeminiVertexExecutor) ExecuteStream(ctx context.Context, auth *cliproxy
}
}()
scanner := bufio.NewScanner(httpResp.Body)
buf := make([]byte, 20_971_520)
scanner.Buffer(buf, 20_971_520)
scanner.Buffer(nil, 20_971_520)
var param any
for scanner.Scan() {
line := scanner.Bytes()

View File

@@ -112,6 +112,8 @@ func (e *IFlowExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, re
}
appendAPIResponseChunk(ctx, e.cfg, data)
reporter.publish(ctx, parseOpenAIUsage(data))
// Ensure usage is recorded even if upstream omits usage metadata.
reporter.ensurePublished(ctx)
var param any
out := sdktranslator.TranslateNonStream(ctx, to, from, req.Model, bytes.Clone(opts.OriginalRequest), body, data, &param)
@@ -199,8 +201,7 @@ func (e *IFlowExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Au
}()
scanner := bufio.NewScanner(httpResp.Body)
buf := make([]byte, 20_971_520)
scanner.Buffer(buf, 20_971_520)
scanner.Buffer(nil, 20_971_520)
var param any
for scanner.Scan() {
line := scanner.Bytes()
@@ -218,6 +219,8 @@ func (e *IFlowExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Au
reporter.publishFailure(ctx)
out <- cliproxyexecutor.StreamChunk{Err: errScan}
}
// Guarantee a usage record exists even if the stream never emitted usage data.
reporter.ensurePublished(ctx)
}()
return stream, nil
@@ -243,13 +246,87 @@ func (e *IFlowExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Auth
return cliproxyexecutor.Response{Payload: []byte(translated)}, nil
}
// Refresh refreshes OAuth tokens and updates the stored API key.
// Refresh refreshes OAuth tokens or cookie-based API keys and updates the stored API key.
func (e *IFlowExecutor) Refresh(ctx context.Context, auth *cliproxyauth.Auth) (*cliproxyauth.Auth, error) {
log.Debugf("iflow executor: refresh called")
if auth == nil {
return nil, fmt.Errorf("iflow executor: auth is nil")
}
// Check if this is cookie-based authentication
var cookie string
var email string
if auth.Metadata != nil {
if v, ok := auth.Metadata["cookie"].(string); ok {
cookie = strings.TrimSpace(v)
}
if v, ok := auth.Metadata["email"].(string); ok {
email = strings.TrimSpace(v)
}
}
// If cookie is present, use cookie-based refresh
if cookie != "" && email != "" {
return e.refreshCookieBased(ctx, auth, cookie, email)
}
// Otherwise, use OAuth-based refresh
return e.refreshOAuthBased(ctx, auth)
}
// refreshCookieBased refreshes API key using browser cookie
func (e *IFlowExecutor) refreshCookieBased(ctx context.Context, auth *cliproxyauth.Auth, cookie, email string) (*cliproxyauth.Auth, error) {
log.Debugf("iflow executor: checking refresh need for cookie-based API key for user: %s", email)
// Get current expiry time from metadata
var currentExpire string
if auth.Metadata != nil {
if v, ok := auth.Metadata["expired"].(string); ok {
currentExpire = strings.TrimSpace(v)
}
}
// Check if refresh is needed
needsRefresh, _, err := iflowauth.ShouldRefreshAPIKey(currentExpire)
if err != nil {
log.Warnf("iflow executor: failed to check refresh need: %v", err)
// If we can't check, continue with refresh anyway as a safety measure
} else if !needsRefresh {
log.Debugf("iflow executor: no refresh needed for user: %s", email)
return auth, nil
}
log.Infof("iflow executor: refreshing cookie-based API key for user: %s", email)
svc := iflowauth.NewIFlowAuth(e.cfg)
keyData, err := svc.RefreshAPIKey(ctx, cookie, email)
if err != nil {
log.Errorf("iflow executor: cookie-based API key refresh failed: %v", err)
return nil, err
}
if auth.Metadata == nil {
auth.Metadata = make(map[string]any)
}
auth.Metadata["api_key"] = keyData.APIKey
auth.Metadata["expired"] = keyData.ExpireTime
auth.Metadata["type"] = "iflow"
auth.Metadata["last_refresh"] = time.Now().Format(time.RFC3339)
auth.Metadata["cookie"] = cookie
auth.Metadata["email"] = email
log.Infof("iflow executor: cookie-based API key refreshed successfully, new expiry: %s", keyData.ExpireTime)
if auth.Attributes == nil {
auth.Attributes = make(map[string]string)
}
auth.Attributes["api_key"] = keyData.APIKey
return auth, nil
}
// refreshOAuthBased refreshes tokens using OAuth refresh token
func (e *IFlowExecutor) refreshOAuthBased(ctx context.Context, auth *cliproxyauth.Auth) (*cliproxyauth.Auth, error) {
refreshToken := ""
oldAccessToken := ""
if auth.Metadata != nil {

View File

@@ -323,7 +323,14 @@ func formatAuthInfo(info upstreamRequestLog) string {
}
func summarizeErrorBody(contentType string, body []byte) string {
if strings.Contains(strings.ToLower(contentType), "text/html") {
isHTML := strings.Contains(strings.ToLower(contentType), "text/html")
if !isHTML {
trimmed := bytes.TrimSpace(bytes.ToLower(body))
if bytes.HasPrefix(trimmed, []byte("<!doctype html")) || bytes.HasPrefix(trimmed, []byte("<html")) {
isHTML = true
}
}
if isHTML {
if title := extractHTMLTitle(body); title != "" {
return title
}

View File

@@ -8,6 +8,7 @@ import (
"io"
"net/http"
"strings"
"time"
"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
@@ -205,8 +206,7 @@ func (e *OpenAICompatExecutor) ExecuteStream(ctx context.Context, auth *cliproxy
}
}()
scanner := bufio.NewScanner(httpResp.Body)
buf := make([]byte, 20_971_520)
scanner.Buffer(buf, 20_971_520)
scanner.Buffer(nil, 20_971_520)
var param any
for scanner.Scan() {
line := scanner.Bytes()
@@ -341,8 +341,9 @@ func (e *OpenAICompatExecutor) overrideModel(payload []byte, model string) []byt
}
type statusErr struct {
code int
msg string
code int
msg string
retryAfter *time.Duration
}
func (e statusErr) Error() string {
@@ -351,4 +352,5 @@ func (e statusErr) Error() string {
}
return fmt.Sprintf("status %d", e.code)
}
func (e statusErr) StatusCode() int { return e.code }
func (e statusErr) StatusCode() int { return e.code }
func (e statusErr) RetryAfter() *time.Duration { return e.retryAfter }

View File

@@ -181,8 +181,7 @@ func (e *QwenExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Aut
}
}()
scanner := bufio.NewScanner(httpResp.Body)
buf := make([]byte, 20_971_520)
scanner.Buffer(buf, 20_971_520)
scanner.Buffer(nil, 20_971_520)
var param any
for scanner.Scan() {
line := scanner.Bytes()

View File

@@ -12,12 +12,14 @@ import (
cliproxyauth "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/auth"
"github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/usage"
"github.com/tidwall/gjson"
"github.com/tidwall/sjson"
)
type usageReporter struct {
provider string
model string
authID string
authIndex uint64
apiKey string
source string
requestedAt time.Time
@@ -35,6 +37,7 @@ func newUsageReporter(ctx context.Context, provider, model string, auth *cliprox
}
if auth != nil {
reporter.authID = auth.ID
reporter.authIndex = auth.Index
}
return reporter
}
@@ -76,6 +79,7 @@ func (r *usageReporter) publishWithOutcome(ctx context.Context, detail usage.Det
Source: r.source,
APIKey: r.apiKey,
AuthID: r.authID,
AuthIndex: r.authIndex,
RequestedAt: r.requestedAt,
Failed: failed,
Detail: detail,
@@ -98,6 +102,7 @@ func (r *usageReporter) ensurePublished(ctx context.Context) {
Source: r.source,
APIKey: r.apiKey,
AuthID: r.authID,
AuthIndex: r.authIndex,
RequestedAt: r.requestedAt,
Failed: false,
Detail: usage.Detail{},
@@ -379,3 +384,94 @@ func jsonPayload(line []byte) []byte {
}
return trimmed
}
// FilterSSEUsageMetadata removes usageMetadata from intermediate SSE events so that
// only the terminal chunk retains token statistics.
// This function is shared between aistudio and antigravity executors.
func FilterSSEUsageMetadata(payload []byte) []byte {
if len(payload) == 0 {
return payload
}
lines := bytes.Split(payload, []byte("\n"))
modified := false
for idx, line := range lines {
trimmed := bytes.TrimSpace(line)
if len(trimmed) == 0 || !bytes.HasPrefix(trimmed, []byte("data:")) {
continue
}
dataIdx := bytes.Index(line, []byte("data:"))
if dataIdx < 0 {
continue
}
rawJSON := bytes.TrimSpace(line[dataIdx+5:])
cleaned, changed := StripUsageMetadataFromJSON(rawJSON)
if !changed {
continue
}
var rebuilt []byte
rebuilt = append(rebuilt, line[:dataIdx]...)
rebuilt = append(rebuilt, []byte("data:")...)
if len(cleaned) > 0 {
rebuilt = append(rebuilt, ' ')
rebuilt = append(rebuilt, cleaned...)
}
lines[idx] = rebuilt
modified = true
}
if !modified {
return payload
}
return bytes.Join(lines, []byte("\n"))
}
// StripUsageMetadataFromJSON drops usageMetadata when no finishReason is present.
// This function is shared between aistudio and antigravity executors.
// It handles both formats:
// - Aistudio: candidates.0.finishReason
// - Antigravity: response.candidates.0.finishReason
func StripUsageMetadataFromJSON(rawJSON []byte) ([]byte, bool) {
jsonBytes := bytes.TrimSpace(rawJSON)
if len(jsonBytes) == 0 || !gjson.ValidBytes(jsonBytes) {
return rawJSON, false
}
// Check for finishReason in both aistudio and antigravity formats
finishReason := gjson.GetBytes(jsonBytes, "candidates.0.finishReason")
if !finishReason.Exists() {
finishReason = gjson.GetBytes(jsonBytes, "response.candidates.0.finishReason")
}
// If finishReason exists and is not empty, keep the usageMetadata
if finishReason.Exists() && finishReason.String() != "" {
return rawJSON, false
}
// Check for usageMetadata in both possible locations
usageMetadata := gjson.GetBytes(jsonBytes, "usageMetadata")
if !usageMetadata.Exists() {
usageMetadata = gjson.GetBytes(jsonBytes, "response.usageMetadata")
}
if !usageMetadata.Exists() {
return rawJSON, false
}
// Remove usageMetadata from both possible locations
cleaned := jsonBytes
var changed bool
// Try to remove usageMetadata from root level
if gjson.GetBytes(cleaned, "usageMetadata").Exists() {
cleaned, _ = sjson.DeleteBytes(cleaned, "usageMetadata")
changed = true
}
// Try to remove usageMetadata from response level
if gjson.GetBytes(cleaned, "response.usageMetadata").Exists() {
cleaned, _ = sjson.DeleteBytes(cleaned, "response.usageMetadata")
changed = true
}
return cleaned, changed
}

View File

@@ -289,7 +289,7 @@ func buildShortNameMap(names []string) map[string]string {
}
base := cand
for i := 1; ; i++ {
suffix := "~" + strconv.Itoa(i)
suffix := "_" + strconv.Itoa(i)
allowed := limit - len(suffix)
if allowed < 0 {
allowed = 0

View File

@@ -310,7 +310,7 @@ func buildShortNameMap(names []string) map[string]string {
}
base := cand
for i := 1; ; i++ {
suffix := "~" + strconv.Itoa(i)
suffix := "_" + strconv.Itoa(i)
allowed := limit - len(suffix)
if allowed < 0 {
allowed = 0

View File

@@ -361,7 +361,7 @@ func buildShortNameMap(names []string) map[string]string {
}
base := cand
for i := 1; ; i++ {
suffix := "~" + strconv.Itoa(i)
suffix := "_" + strconv.Itoa(i)
allowed := limit - len(suffix)
if allowed < 0 {
allowed = 0

View File

@@ -22,6 +22,7 @@ func ConvertOpenAIResponsesRequestToCodex(modelName string, inputRawJSON []byte,
rawJSON, _ = sjson.DeleteBytes(rawJSON, "max_completion_tokens")
rawJSON, _ = sjson.DeleteBytes(rawJSON, "temperature")
rawJSON, _ = sjson.DeleteBytes(rawJSON, "top_p")
rawJSON, _ = sjson.DeleteBytes(rawJSON, "service_tier")
originalInstructions := ""
originalInstructionsText := ""

View File

@@ -17,6 +17,8 @@ import (
"github.com/tidwall/sjson"
)
const geminiCLIClaudeThoughtSignature = "skip_thought_signature_validator"
// ConvertClaudeRequestToCLI parses and transforms a Claude Code API request into Gemini CLI API format.
// It extracts the model name, system instruction, message contents, and tool declarations
// from the raw JSON request and returns them in the format expected by the Gemini CLI API.
@@ -89,7 +91,10 @@ func ConvertClaudeRequestToCLI(modelName string, inputRawJSON []byte, _ bool) []
functionArgs := contentResult.Get("input").String()
var args map[string]any
if err := json.Unmarshal([]byte(functionArgs), &args); err == nil {
clientContent.Parts = append(clientContent.Parts, client.Part{FunctionCall: &client.FunctionCall{Name: functionName, Args: args}})
clientContent.Parts = append(clientContent.Parts, client.Part{
FunctionCall: &client.FunctionCall{Name: functionName, Args: args},
ThoughtSignature: geminiCLIClaudeThoughtSignature,
})
}
} else if contentTypeResult.Type == gjson.String && contentTypeResult.String() == "tool_result" {
toolCallID := contentResult.Get("tool_use_id").String()
@@ -128,6 +133,7 @@ func ConvertClaudeRequestToCLI(modelName string, inputRawJSON []byte, _ bool) []
tool, _ := sjson.Delete(toolResult.Raw, "input_schema")
tool, _ = sjson.SetRaw(tool, "parametersJsonSchema", inputSchema)
tool, _ = sjson.Delete(tool, "strict")
tool, _ = sjson.Delete(tool, "input_examples")
var toolDeclaration any
if err := json.Unmarshal([]byte(tool), &toolDeclaration); err == nil {
tools[0].FunctionDeclarations = append(tools[0].FunctionDeclarations, toolDeclaration)

View File

@@ -6,6 +6,7 @@
package gemini
import (
"bytes"
"context"
"fmt"
@@ -13,7 +14,7 @@ import (
"github.com/tidwall/sjson"
)
// ConvertGeminiCliRequestToGemini parses and transforms a Gemini CLI API request into Gemini API format.
// ConvertGeminiCliResponseToGemini parses and transforms a Gemini CLI API request into Gemini API format.
// It extracts the model name, system instruction, message contents, and tool declarations
// from the raw JSON request and returns them in the format expected by the Gemini API.
// The function performs the following transformations:
@@ -29,7 +30,11 @@ import (
//
// Returns:
// - []string: The transformed request data in Gemini API format
func ConvertGeminiCliRequestToGemini(ctx context.Context, _ string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, _ *any) []string {
func ConvertGeminiCliResponseToGemini(ctx context.Context, _ string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, _ *any) []string {
if bytes.HasPrefix(rawJSON, []byte("data:")) {
rawJSON = bytes.TrimSpace(rawJSON[5:])
}
if alt, ok := ctx.Value("alt").(string); ok {
var chunk []byte
if alt == "" {
@@ -56,7 +61,7 @@ func ConvertGeminiCliRequestToGemini(ctx context.Context, _ string, originalRequ
return []string{}
}
// ConvertGeminiCliRequestToGeminiNonStream converts a non-streaming Gemini CLI request to a non-streaming Gemini response.
// ConvertGeminiCliResponseToGeminiNonStream converts a non-streaming Gemini CLI request to a non-streaming Gemini response.
// This function processes the complete Gemini CLI request and transforms it into a single Gemini-compatible
// JSON response. It extracts the response data from the request and returns it in the expected format.
//
@@ -68,7 +73,7 @@ func ConvertGeminiCliRequestToGemini(ctx context.Context, _ string, originalRequ
//
// Returns:
// - string: A Gemini-compatible JSON response containing the response data
func ConvertGeminiCliRequestToGeminiNonStream(_ context.Context, _ string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, _ *any) string {
func ConvertGeminiCliResponseToGeminiNonStream(_ context.Context, _ string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, _ *any) string {
responseResult := gjson.GetBytes(rawJSON, "response")
if responseResult.Exists() {
return responseResult.Raw

View File

@@ -12,8 +12,8 @@ func init() {
GeminiCLI,
ConvertGeminiRequestToGeminiCLI,
interfaces.TranslateResponse{
Stream: ConvertGeminiCliRequestToGemini,
NonStream: ConvertGeminiCliRequestToGeminiNonStream,
Stream: ConvertGeminiCliResponseToGemini,
NonStream: ConvertGeminiCliResponseToGeminiNonStream,
TokenCount: GeminiTokenCount,
},
)

View File

@@ -15,6 +15,8 @@ import (
"github.com/tidwall/sjson"
)
const geminiCLIFunctionThoughtSignature = "skip_thought_signature_validator"
// ConvertOpenAIRequestToGeminiCLI converts an OpenAI Chat Completions request (raw JSON)
// into a complete Gemini CLI request JSON. All JSON construction uses sjson and lookups use gjson.
//
@@ -86,6 +88,15 @@ func ConvertOpenAIRequestToGeminiCLI(modelName string, inputRawJSON []byte, _ bo
}
}
// For gemini-3-pro-preview, always send default thinkingConfig when none specified.
// This matches the official Gemini CLI behavior which always sends:
// { thinkingBudget: -1, includeThoughts: true }
// See: ai-gemini-cli/packages/core/src/config/defaultModelConfigs.ts
if !gjson.GetBytes(out, "request.generationConfig.thinkingConfig").Exists() && modelName == "gemini-3-pro-preview" {
out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", -1)
out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
}
// Temperature/top_p/top_k
if tr := gjson.GetBytes(rawJSON, "temperature"); tr.Exists() && tr.Type == gjson.Number {
out, _ = sjson.SetBytes(out, "request.generationConfig.temperature", tr.Num)
@@ -239,6 +250,7 @@ func ConvertOpenAIRequestToGeminiCLI(modelName string, inputRawJSON []byte, _ bo
fargs := tc.Get("function.arguments").String()
node, _ = sjson.SetBytes(node, "parts."+itoa(p)+".functionCall.name", fname)
node, _ = sjson.SetRawBytes(node, "parts."+itoa(p)+".functionCall.args", []byte(fargs))
node, _ = sjson.SetBytes(node, "parts."+itoa(p)+".thoughtSignature", geminiCLIFunctionThoughtSignature)
p++
if fid != "" {
fIDs = append(fIDs, fid)
@@ -283,6 +295,17 @@ func ConvertOpenAIRequestToGeminiCLI(modelName string, inputRawJSON []byte, _ bo
renamed, errRename := util.RenameKey(fnRaw, "parameters", "parametersJsonSchema")
if errRename != nil {
log.Warnf("Failed to rename parameters for tool '%s': %v", fn.Get("name").String(), errRename)
var errSet error
fnRaw, errSet = sjson.Set(fnRaw, "parametersJsonSchema.type", "object")
if errSet != nil {
log.Warnf("Failed to set default schema type for tool '%s': %v", fn.Get("name").String(), errSet)
continue
}
fnRaw, errSet = sjson.Set(fnRaw, "parametersJsonSchema.properties", map[string]interface{}{})
if errSet != nil {
log.Warnf("Failed to set default schema properties for tool '%s': %v", fn.Get("name").String(), errSet)
continue
}
} else {
fnRaw = renamed
}

View File

@@ -98,25 +98,40 @@ func ConvertCliResponseToOpenAI(_ context.Context, _ string, originalRequestRawJ
// Process the main content part of the response.
partsResult := gjson.GetBytes(rawJSON, "response.candidates.0.content.parts")
hasFunctionCall := false
hasValidContent := false
if partsResult.IsArray() {
partResults := partsResult.Array()
for i := 0; i < len(partResults); i++ {
partResult := partResults[i]
partTextResult := partResult.Get("text")
functionCallResult := partResult.Get("functionCall")
thoughtSignatureResult := partResult.Get("thoughtSignature")
inlineDataResult := partResult.Get("inlineData")
if !inlineDataResult.Exists() {
inlineDataResult = partResult.Get("inline_data")
}
// Handle thoughtSignature - this is encrypted reasoning content that should not be exposed to the client
if thoughtSignatureResult.Exists() && thoughtSignatureResult.String() != "" {
// Skip thoughtSignature processing - it's internal encrypted data
continue
}
if partTextResult.Exists() {
textContent := partTextResult.String()
// Skip empty text content to avoid generating unnecessary chunks
if textContent == "" {
continue
}
// Handle text content, distinguishing between regular content and reasoning/thoughts.
if partResult.Get("thought").Bool() {
template, _ = sjson.Set(template, "choices.0.delta.reasoning_content", partTextResult.String())
template, _ = sjson.Set(template, "choices.0.delta.reasoning_content", textContent)
} else {
template, _ = sjson.Set(template, "choices.0.delta.content", partTextResult.String())
template, _ = sjson.Set(template, "choices.0.delta.content", textContent)
}
template, _ = sjson.Set(template, "choices.0.delta.role", "assistant")
hasValidContent = true
} else if functionCallResult.Exists() {
// Handle function call content.
hasFunctionCall = true
@@ -176,6 +191,12 @@ func ConvertCliResponseToOpenAI(_ context.Context, _ string, originalRequestRawJ
template, _ = sjson.Set(template, "choices.0.native_finish_reason", "tool_calls")
}
// Only return a chunk if there's actual content or a finish reason
finishReason := gjson.GetBytes(rawJSON, "response.candidates.0.finishReason")
if !hasValidContent && !finishReason.Exists() {
return []string{}
}
return []string{template}
}

View File

@@ -17,6 +17,8 @@ import (
"github.com/tidwall/sjson"
)
const geminiClaudeThoughtSignature = "skip_thought_signature_validator"
// ConvertClaudeRequestToGemini parses a Claude API request and returns a complete
// Gemini CLI request body (as JSON bytes) ready to be sent via SendRawMessageStream.
// All JSON transformations are performed using gjson/sjson.
@@ -82,7 +84,10 @@ func ConvertClaudeRequestToGemini(modelName string, inputRawJSON []byte, _ bool)
functionArgs := contentResult.Get("input").String()
var args map[string]any
if err := json.Unmarshal([]byte(functionArgs), &args); err == nil {
clientContent.Parts = append(clientContent.Parts, client.Part{FunctionCall: &client.FunctionCall{Name: functionName, Args: args}})
clientContent.Parts = append(clientContent.Parts, client.Part{
FunctionCall: &client.FunctionCall{Name: functionName, Args: args},
ThoughtSignature: geminiClaudeThoughtSignature,
})
}
} else if contentTypeResult.Type == gjson.String && contentTypeResult.String() == "tool_result" {
toolCallID := contentResult.Get("tool_use_id").String()
@@ -121,6 +126,7 @@ func ConvertClaudeRequestToGemini(modelName string, inputRawJSON []byte, _ bool)
tool, _ := sjson.Delete(toolResult.Raw, "input_schema")
tool, _ = sjson.SetRaw(tool, "parametersJsonSchema", inputSchema)
tool, _ = sjson.Delete(tool, "strict")
tool, _ = sjson.Delete(tool, "input_examples")
var toolDeclaration any
if err := json.Unmarshal([]byte(tool), &toolDeclaration); err == nil {
tools[0].FunctionDeclarations = append(tools[0].FunctionDeclarations, toolDeclaration)

View File

@@ -15,6 +15,8 @@ import (
"github.com/tidwall/sjson"
)
const geminiFunctionThoughtSignature = "skip_thought_signature_validator"
// ConvertOpenAIRequestToGemini converts an OpenAI Chat Completions request (raw JSON)
// into a complete Gemini request JSON. All JSON construction uses sjson and lookups use gjson.
//
@@ -264,6 +266,7 @@ func ConvertOpenAIRequestToGemini(modelName string, inputRawJSON []byte, _ bool)
fargs := tc.Get("function.arguments").String()
node, _ = sjson.SetBytes(node, "parts."+itoa(p)+".functionCall.name", fname)
node, _ = sjson.SetRawBytes(node, "parts."+itoa(p)+".functionCall.args", []byte(fargs))
node, _ = sjson.SetBytes(node, "parts."+itoa(p)+".thoughtSignature", geminiFunctionThoughtSignature)
p++
if fid != "" {
fIDs = append(fIDs, fid)
@@ -303,8 +306,34 @@ func ConvertOpenAIRequestToGemini(modelName string, inputRawJSON []byte, _ bool)
if t.Get("type").String() == "function" {
fn := t.Get("function")
if fn.Exists() && fn.IsObject() {
parametersJsonSchema, _ := util.RenameKey(fn.Raw, "parameters", "parametersJsonSchema")
out, _ = sjson.SetRawBytes(out, fdPath+".-1", []byte(parametersJsonSchema))
fnRaw := fn.Raw
if fn.Get("parameters").Exists() {
renamed, errRename := util.RenameKey(fnRaw, "parameters", "parametersJsonSchema")
if errRename != nil {
log.Warnf("Failed to rename parameters for tool '%s': %v", fn.Get("name").String(), errRename)
} else {
fnRaw = renamed
}
} else {
var errSet error
fnRaw, errSet = sjson.Set(fnRaw, "parametersJsonSchema.type", "object")
if errSet != nil {
log.Warnf("Failed to set default schema type for tool '%s': %v", fn.Get("name").String(), errSet)
continue
}
fnRaw, errSet = sjson.Set(fnRaw, "parametersJsonSchema.properties", map[string]interface{}{})
if errSet != nil {
log.Warnf("Failed to set default schema properties for tool '%s': %v", fn.Get("name").String(), errSet)
continue
}
}
fnRaw, _ = sjson.Delete(fnRaw, "strict")
tmp, errSet := sjson.SetRawBytes(out, fdPath+".-1", []byte(fnRaw))
if errSet != nil {
log.Warnf("Failed to append tool declaration for '%s': %v", fn.Get("name").String(), errSet)
continue
}
out = tmp
}
}
}

View File

@@ -10,6 +10,8 @@ import (
"github.com/tidwall/sjson"
)
const geminiResponsesThoughtSignature = "skip_thought_signature_validator"
func ConvertOpenAIResponsesRequestToGemini(modelName string, inputRawJSON []byte, stream bool) []byte {
rawJSON := bytes.Clone(inputRawJSON)
@@ -108,6 +110,7 @@ func ConvertOpenAIResponsesRequestToGemini(modelName string, inputRawJSON []byte
modelContent := `{"role":"model","parts":[]}`
functionCall := `{"functionCall":{"name":"","args":{}}}`
functionCall, _ = sjson.Set(functionCall, "functionCall.name", name)
functionCall, _ = sjson.Set(functionCall, "thoughtSignature", geminiResponsesThoughtSignature)
// Parse arguments JSON string and set as args object
if arguments != "" {
@@ -291,6 +294,17 @@ func ConvertOpenAIResponsesRequestToGemini(modelName string, inputRawJSON []byte
}
}
}
// For gemini-3-pro-preview, always send default thinkingConfig when none specified.
// This matches the official Gemini CLI behavior which always sends:
// { thinkingBudget: -1, includeThoughts: true }
// See: ai-gemini-cli/packages/core/src/config/defaultModelConfigs.ts
if !gjson.Get(out, "generationConfig.thinkingConfig").Exists() && modelName == "gemini-3-pro-preview" {
out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", -1)
out, _ = sjson.Set(out, "generationConfig.thinkingConfig.include_thoughts", true)
// log.Debugf("Applied default thinkingConfig for gemini-3-pro-preview (matches Gemini CLI): thinkingBudget=-1, include_thoughts=true")
}
result := []byte(out)
result = common.AttachDefaultSafetySettings(result, "safetySettings")
return result

View File

@@ -691,14 +691,10 @@ func extractReasoningTexts(node gjson.Result) []string {
switch node.Type {
case gjson.String:
if text := strings.TrimSpace(node.String()); text != "" {
texts = append(texts, text)
}
texts = append(texts, node.String())
case gjson.JSON:
if text := node.Get("text"); text.Exists() {
if trimmed := strings.TrimSpace(text.String()); trimmed != "" {
texts = append(texts, trimmed)
}
texts = append(texts, text.String())
} else if raw := strings.TrimSpace(node.Raw); raw != "" && !strings.HasPrefix(raw, "{") && !strings.HasPrefix(raw, "[") {
texts = append(texts, raw)
}

View File

@@ -90,6 +90,7 @@ type modelStats struct {
type RequestDetail struct {
Timestamp time.Time `json:"timestamp"`
Source string `json:"source"`
AuthIndex uint64 `json:"auth_index"`
Tokens TokenStats `json:"tokens"`
Failed bool `json:"failed"`
}
@@ -197,6 +198,7 @@ func (s *RequestStatistics) Record(ctx context.Context, record coreusage.Record)
s.updateAPIStats(stats, modelName, RequestDetail{
Timestamp: timestamp,
Source: record.Source,
AuthIndex: record.AuthIndex,
Tokens: detail,
Failed: failed,
})

View File

@@ -5,6 +5,7 @@ import (
"strconv"
"strings"
"github.com/tidwall/gjson"
"github.com/tidwall/sjson"
)
@@ -62,6 +63,23 @@ func ParseGeminiThinkingSuffix(model string) (string, *int, *bool, bool) {
return base, &budgetValue, nil, true
}
func NormalizeGeminiThinkingModel(modelName string) (string, map[string]any) {
baseModel, budget, include, matched := ParseGeminiThinkingSuffix(modelName)
if !matched {
return baseModel, nil
}
metadata := map[string]any{
GeminiOriginalModelMetadataKey: modelName,
}
if budget != nil {
metadata[GeminiThinkingBudgetMetadataKey] = *budget
}
if include != nil {
metadata[GeminiIncludeThoughtsMetadataKey] = *include
}
return baseModel, metadata
}
func ApplyGeminiThinkingConfig(body []byte, budget *int, includeThoughts *bool) []byte {
if budget == nil && includeThoughts == nil {
return body
@@ -195,3 +213,49 @@ func StripThinkingConfigIfUnsupported(model string, body []byte) []byte {
updated, _ = sjson.DeleteBytes(updated, "generationConfig.thinkingConfig")
return updated
}
// ConvertThinkingLevelToBudget checks for "generationConfig.thinkingConfig.thinkingLevel"
// and converts it to "thinkingBudget".
// "high" -> 32768
// "low" -> 128
// It removes "thinkingLevel" after conversion.
func ConvertThinkingLevelToBudget(body []byte) []byte {
levelPath := "generationConfig.thinkingConfig.thinkingLevel"
res := gjson.GetBytes(body, levelPath)
if !res.Exists() {
return body
}
level := strings.ToLower(res.String())
var budget int
switch level {
case "high":
budget = 32768
case "low":
budget = 128
default:
// If unknown level, we might just leave it or default.
// User only specified high and low. We'll assume we shouldn't touch it if it's something else,
// or maybe we should just remove the invalid level?
// For safety adhering to strict instructions: "If high... if low...".
// If it's something else, the upstream might fail anyway if we leave it,
// but let's just delete the level if we processed it.
// Actually, let's check if we need to do anything for other values.
// For now, only handle high/low.
return body
}
// Set budget
budgetPath := "generationConfig.thinkingConfig.thinkingBudget"
updated, err := sjson.SetBytes(body, budgetPath, budget)
if err != nil {
return body
}
// Remove level
updated, err = sjson.DeleteBytes(updated, levelPath)
if err != nil {
return body
}
return updated
}

View File

@@ -8,9 +8,12 @@ package claude
import (
"bufio"
"bytes"
"compress/gzip"
"context"
"encoding/json"
"fmt"
"io"
"net/http"
"time"
@@ -19,6 +22,7 @@ import (
"github.com/router-for-me/CLIProxyAPI/v6/internal/interfaces"
"github.com/router-for-me/CLIProxyAPI/v6/internal/registry"
"github.com/router-for-me/CLIProxyAPI/v6/sdk/api/handlers"
log "github.com/sirupsen/logrus"
"github.com/tidwall/gjson"
)
@@ -153,6 +157,23 @@ func (h *ClaudeCodeAPIHandler) handleNonStreamingResponse(c *gin.Context, rawJSO
cliCancel(errMsg.Error)
return
}
// Decompress gzipped responses - Claude API sometimes returns gzip without Content-Encoding header
// This fixes title generation and other non-streaming responses that arrive compressed
if len(resp) >= 2 && resp[0] == 0x1f && resp[1] == 0x8b {
gzReader, err := gzip.NewReader(bytes.NewReader(resp))
if err != nil {
log.Warnf("failed to decompress gzipped Claude response: %v", err)
} else {
defer gzReader.Close()
if decompressed, err := io.ReadAll(gzReader); err != nil {
log.Warnf("failed to read decompressed Claude response: %v", err)
} else {
resp = decompressed
}
}
}
_, _ = c.Writer.Write(resp)
cliCancel()
}

View File

@@ -359,20 +359,7 @@ func cloneBytes(src []byte) []byte {
}
func normalizeModelMetadata(modelName string) (string, map[string]any) {
baseModel, budget, include, matched := util.ParseGeminiThinkingSuffix(modelName)
if !matched {
return baseModel, nil
}
metadata := map[string]any{
util.GeminiOriginalModelMetadataKey: modelName,
}
if budget != nil {
metadata[util.GeminiThinkingBudgetMetadataKey] = *budget
}
if include != nil {
metadata[util.GeminiIncludeThoughtsMetadataKey] = *include
}
return baseModel, metadata
return util.NormalizeGeminiThinkingModel(modelName)
}
func cloneMetadata(src map[string]any) map[string]any {

293
sdk/auth/antigravity.go Normal file
View File

@@ -0,0 +1,293 @@
package auth
import (
"context"
"encoding/json"
"fmt"
"net"
"net/http"
"net/url"
"strings"
"time"
"github.com/router-for-me/CLIProxyAPI/v6/internal/browser"
"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
"github.com/router-for-me/CLIProxyAPI/v6/internal/misc"
"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
coreauth "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/auth"
log "github.com/sirupsen/logrus"
)
const (
antigravityClientID = "1071006060591-tmhssin2h21lcre235vtolojh4g403ep.apps.googleusercontent.com"
antigravityClientSecret = "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"
antigravityCallbackPort = 51121
)
var antigravityScopes = []string{
"https://www.googleapis.com/auth/cloud-platform",
"https://www.googleapis.com/auth/userinfo.email",
"https://www.googleapis.com/auth/userinfo.profile",
"https://www.googleapis.com/auth/cclog",
"https://www.googleapis.com/auth/experimentsandconfigs",
}
// AntigravityAuthenticator implements OAuth login for the antigravity provider.
type AntigravityAuthenticator struct{}
// NewAntigravityAuthenticator constructs a new authenticator instance.
func NewAntigravityAuthenticator() Authenticator { return &AntigravityAuthenticator{} }
// Provider returns the provider key for antigravity.
func (AntigravityAuthenticator) Provider() string { return "antigravity" }
// RefreshLead instructs the manager to refresh five minutes before expiry.
func (AntigravityAuthenticator) RefreshLead() *time.Duration {
lead := 5 * time.Minute
return &lead
}
// Login launches a local OAuth flow to obtain antigravity tokens and persists them.
func (AntigravityAuthenticator) Login(ctx context.Context, cfg *config.Config, opts *LoginOptions) (*coreauth.Auth, error) {
if cfg == nil {
return nil, fmt.Errorf("cliproxy auth: configuration is required")
}
if ctx == nil {
ctx = context.Background()
}
if opts == nil {
opts = &LoginOptions{}
}
httpClient := util.SetProxy(&cfg.SDKConfig, &http.Client{})
state, err := misc.GenerateRandomState()
if err != nil {
return nil, fmt.Errorf("antigravity: failed to generate state: %w", err)
}
srv, port, cbChan, errServer := startAntigravityCallbackServer()
if errServer != nil {
return nil, fmt.Errorf("antigravity: failed to start callback server: %w", errServer)
}
defer func() {
shutdownCtx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
defer cancel()
_ = srv.Shutdown(shutdownCtx)
}()
redirectURI := fmt.Sprintf("http://localhost:%d/oauth-callback", port)
authURL := buildAntigravityAuthURL(redirectURI, state)
if !opts.NoBrowser {
fmt.Println("Opening browser for antigravity authentication")
if !browser.IsAvailable() {
log.Warn("No browser available; please open the URL manually")
util.PrintSSHTunnelInstructions(port)
fmt.Printf("Visit the following URL to continue authentication:\n%s\n", authURL)
} else if errOpen := browser.OpenURL(authURL); errOpen != nil {
log.Warnf("Failed to open browser automatically: %v", errOpen)
util.PrintSSHTunnelInstructions(port)
fmt.Printf("Visit the following URL to continue authentication:\n%s\n", authURL)
}
} else {
util.PrintSSHTunnelInstructions(port)
fmt.Printf("Visit the following URL to continue authentication:\n%s\n", authURL)
}
fmt.Println("Waiting for antigravity authentication callback...")
var cbRes callbackResult
select {
case res := <-cbChan:
cbRes = res
case <-time.After(5 * time.Minute):
return nil, fmt.Errorf("antigravity: authentication timed out")
}
if cbRes.Error != "" {
return nil, fmt.Errorf("antigravity: authentication failed: %s", cbRes.Error)
}
if cbRes.State != state {
return nil, fmt.Errorf("antigravity: invalid state")
}
if cbRes.Code == "" {
return nil, fmt.Errorf("antigravity: missing authorization code")
}
tokenResp, errToken := exchangeAntigravityCode(ctx, cbRes.Code, redirectURI, httpClient)
if errToken != nil {
return nil, fmt.Errorf("antigravity: token exchange failed: %w", errToken)
}
email := ""
if tokenResp.AccessToken != "" {
if info, errInfo := fetchAntigravityUserInfo(ctx, tokenResp.AccessToken, httpClient); errInfo == nil && strings.TrimSpace(info.Email) != "" {
email = strings.TrimSpace(info.Email)
}
}
now := time.Now()
metadata := map[string]any{
"type": "antigravity",
"access_token": tokenResp.AccessToken,
"refresh_token": tokenResp.RefreshToken,
"expires_in": tokenResp.ExpiresIn,
"timestamp": now.UnixMilli(),
"expired": now.Add(time.Duration(tokenResp.ExpiresIn) * time.Second).Format(time.RFC3339),
}
if email != "" {
metadata["email"] = email
}
fileName := sanitizeAntigravityFileName(email)
label := email
if label == "" {
label = "antigravity"
}
fmt.Println("Antigravity authentication successful")
return &coreauth.Auth{
ID: fileName,
Provider: "antigravity",
FileName: fileName,
Label: label,
Metadata: metadata,
}, nil
}
type callbackResult struct {
Code string
Error string
State string
}
func startAntigravityCallbackServer() (*http.Server, int, <-chan callbackResult, error) {
addr := fmt.Sprintf(":%d", antigravityCallbackPort)
listener, err := net.Listen("tcp", addr)
if err != nil {
return nil, 0, nil, err
}
port := listener.Addr().(*net.TCPAddr).Port
resultCh := make(chan callbackResult, 1)
mux := http.NewServeMux()
mux.HandleFunc("/oauth-callback", func(w http.ResponseWriter, r *http.Request) {
q := r.URL.Query()
res := callbackResult{
Code: strings.TrimSpace(q.Get("code")),
Error: strings.TrimSpace(q.Get("error")),
State: strings.TrimSpace(q.Get("state")),
}
resultCh <- res
if res.Code != "" && res.Error == "" {
_, _ = w.Write([]byte("<h1>Login successful</h1><p>You can close this window.</p>"))
} else {
_, _ = w.Write([]byte("<h1>Login failed</h1><p>Please check the CLI output.</p>"))
}
})
srv := &http.Server{Handler: mux}
go func() {
if errServe := srv.Serve(listener); errServe != nil && !strings.Contains(errServe.Error(), "Server closed") {
log.Warnf("antigravity callback server error: %v", errServe)
}
}()
return srv, port, resultCh, nil
}
type antigravityTokenResponse struct {
AccessToken string `json:"access_token"`
RefreshToken string `json:"refresh_token"`
ExpiresIn int64 `json:"expires_in"`
TokenType string `json:"token_type"`
}
func exchangeAntigravityCode(ctx context.Context, code, redirectURI string, httpClient *http.Client) (*antigravityTokenResponse, error) {
data := url.Values{}
data.Set("code", code)
data.Set("client_id", antigravityClientID)
data.Set("client_secret", antigravityClientSecret)
data.Set("redirect_uri", redirectURI)
data.Set("grant_type", "authorization_code")
req, err := http.NewRequestWithContext(ctx, http.MethodPost, "https://oauth2.googleapis.com/token", strings.NewReader(data.Encode()))
if err != nil {
return nil, err
}
req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
resp, errDo := httpClient.Do(req)
if errDo != nil {
return nil, errDo
}
defer func() {
if errClose := resp.Body.Close(); errClose != nil {
log.Errorf("antigravity token exchange: close body error: %v", errClose)
}
}()
var token antigravityTokenResponse
if errDecode := json.NewDecoder(resp.Body).Decode(&token); errDecode != nil {
return nil, errDecode
}
if resp.StatusCode < http.StatusOK || resp.StatusCode >= http.StatusMultipleChoices {
return nil, fmt.Errorf("oauth token exchange failed: status %d", resp.StatusCode)
}
return &token, nil
}
type antigravityUserInfo struct {
Email string `json:"email"`
}
func fetchAntigravityUserInfo(ctx context.Context, accessToken string, httpClient *http.Client) (*antigravityUserInfo, error) {
if strings.TrimSpace(accessToken) == "" {
return &antigravityUserInfo{}, nil
}
req, err := http.NewRequestWithContext(ctx, http.MethodGet, "https://www.googleapis.com/oauth2/v1/userinfo?alt=json", nil)
if err != nil {
return nil, err
}
req.Header.Set("Authorization", "Bearer "+accessToken)
resp, errDo := httpClient.Do(req)
if errDo != nil {
return nil, errDo
}
defer func() {
if errClose := resp.Body.Close(); errClose != nil {
log.Errorf("antigravity userinfo: close body error: %v", errClose)
}
}()
if resp.StatusCode < http.StatusOK || resp.StatusCode >= http.StatusMultipleChoices {
return &antigravityUserInfo{}, nil
}
var info antigravityUserInfo
if errDecode := json.NewDecoder(resp.Body).Decode(&info); errDecode != nil {
return nil, errDecode
}
return &info, nil
}
func buildAntigravityAuthURL(redirectURI, state string) string {
params := url.Values{}
params.Set("access_type", "offline")
params.Set("client_id", antigravityClientID)
params.Set("prompt", "consent")
params.Set("redirect_uri", redirectURI)
params.Set("response_type", "code")
params.Set("scope", strings.Join(antigravityScopes, " "))
params.Set("state", state)
return "https://accounts.google.com/o/oauth2/v2/auth?" + params.Encode()
}
func sanitizeAntigravityFileName(email string) string {
if strings.TrimSpace(email) == "" {
return "antigravity.json"
}
replacer := strings.NewReplacer("@", "_", ".", "_")
return fmt.Sprintf("antigravity-%s.json", replacer.Replace(email))
}

View File

@@ -13,6 +13,7 @@ func init() {
registerRefreshLead("iflow", func() Authenticator { return NewIFlowAuthenticator() })
registerRefreshLead("gemini", func() Authenticator { return NewGeminiAuthenticator() })
registerRefreshLead("gemini-cli", func() Authenticator { return NewGeminiAuthenticator() })
registerRefreshLead("antigravity", func() Authenticator { return NewAntigravityAuthenticator() })
}
func registerRefreshLead(provider string, factory func() Authenticator) {

View File

@@ -62,6 +62,8 @@ type Result struct {
Model string
// Success marks whether the execution succeeded.
Success bool
// RetryAfter carries a provider supplied retry hint (e.g. 429 retryDelay).
RetryAfter *time.Duration
// Error describes the failure when Success is false.
Error *Error
}
@@ -169,6 +171,7 @@ func (m *Manager) Register(ctx context.Context, auth *Auth) (*Auth, error) {
if auth == nil {
return nil, nil
}
auth.EnsureIndex()
if auth.ID == "" {
auth.ID = uuid.NewString()
}
@@ -185,6 +188,7 @@ func (m *Manager) Update(ctx context.Context, auth *Auth) (*Auth, error) {
if auth == nil || auth.ID == "" {
return nil, nil
}
auth.EnsureIndex()
m.mu.Lock()
m.auths[auth.ID] = auth.Clone()
m.mu.Unlock()
@@ -209,6 +213,7 @@ func (m *Manager) Load(ctx context.Context) error {
if auth == nil || auth.ID == "" {
continue
}
auth.EnsureIndex()
m.auths[auth.ID] = auth.Clone()
}
return nil
@@ -322,6 +327,9 @@ func (m *Manager) executeWithProvider(ctx context.Context, provider string, req
if errors.As(errExec, &se) && se != nil {
result.Error.HTTPStatus = se.StatusCode()
}
if ra := retryAfterFromError(errExec); ra != nil {
result.RetryAfter = ra
}
m.MarkResult(execCtx, result)
lastErr = errExec
continue
@@ -367,6 +375,9 @@ func (m *Manager) executeCountWithProvider(ctx context.Context, provider string,
if errors.As(errExec, &se) && se != nil {
result.Error.HTTPStatus = se.StatusCode()
}
if ra := retryAfterFromError(errExec); ra != nil {
result.RetryAfter = ra
}
m.MarkResult(execCtx, result)
lastErr = errExec
continue
@@ -412,6 +423,7 @@ func (m *Manager) executeStreamWithProvider(ctx context.Context, provider string
rerr.HTTPStatus = se.StatusCode()
}
result := Result{AuthID: auth.ID, Provider: provider, Model: req.Model, Success: false, Error: rerr}
result.RetryAfter = retryAfterFromError(errStream)
m.MarkResult(execCtx, result)
lastErr = errStream
continue
@@ -552,18 +564,29 @@ func (m *Manager) MarkResult(ctx context.Context, result Result) {
state.NextRetryAfter = next
suspendReason = "payment_required"
shouldSuspendModel = true
case 404:
next := now.Add(12 * time.Hour)
state.NextRetryAfter = next
suspendReason = "not_found"
shouldSuspendModel = true
case 429:
cooldown, nextLevel := nextQuotaCooldown(state.Quota.BackoffLevel)
var next time.Time
if cooldown > 0 {
next = now.Add(cooldown)
backoffLevel := state.Quota.BackoffLevel
if result.RetryAfter != nil {
next = now.Add(*result.RetryAfter)
} else {
cooldown, nextLevel := nextQuotaCooldown(backoffLevel)
if cooldown > 0 {
next = now.Add(cooldown)
}
backoffLevel = nextLevel
}
state.NextRetryAfter = next
state.Quota = QuotaState{
Exceeded: true,
Reason: "quota",
NextRecoverAt: next,
BackoffLevel: nextLevel,
BackoffLevel: backoffLevel,
}
suspendReason = "quota"
shouldSuspendModel = true
@@ -579,7 +602,7 @@ func (m *Manager) MarkResult(ctx context.Context, result Result) {
auth.UpdatedAt = now
updateAggregatedAvailability(auth, now)
} else {
applyAuthFailureState(auth, result.Error, now)
applyAuthFailureState(auth, result.Error, result.RetryAfter, now)
}
}
@@ -739,6 +762,25 @@ func cloneError(err *Error) *Error {
}
}
func retryAfterFromError(err error) *time.Duration {
if err == nil {
return nil
}
type retryAfterProvider interface {
RetryAfter() *time.Duration
}
rap, ok := err.(retryAfterProvider)
if !ok || rap == nil {
return nil
}
retryAfter := rap.RetryAfter()
if retryAfter == nil {
return nil
}
val := *retryAfter
return &val
}
func statusCodeFromResult(err *Error) int {
if err == nil {
return 0
@@ -746,7 +788,7 @@ func statusCodeFromResult(err *Error) int {
return err.StatusCode()
}
func applyAuthFailureState(auth *Auth, resultErr *Error, now time.Time) {
func applyAuthFailureState(auth *Auth, resultErr *Error, retryAfter *time.Duration, now time.Time) {
if auth == nil {
return
}
@@ -767,17 +809,24 @@ func applyAuthFailureState(auth *Auth, resultErr *Error, now time.Time) {
case 402, 403:
auth.StatusMessage = "payment_required"
auth.NextRetryAfter = now.Add(30 * time.Minute)
case 404:
auth.StatusMessage = "not_found"
auth.NextRetryAfter = now.Add(12 * time.Hour)
case 429:
auth.StatusMessage = "quota exhausted"
auth.Quota.Exceeded = true
auth.Quota.Reason = "quota"
cooldown, nextLevel := nextQuotaCooldown(auth.Quota.BackoffLevel)
var next time.Time
if cooldown > 0 {
next = now.Add(cooldown)
if retryAfter != nil {
next = now.Add(*retryAfter)
} else {
cooldown, nextLevel := nextQuotaCooldown(auth.Quota.BackoffLevel)
if cooldown > 0 {
next = now.Add(cooldown)
}
auth.Quota.BackoffLevel = nextLevel
}
auth.Quota.NextRecoverAt = next
auth.Quota.BackoffLevel = nextLevel
auth.NextRetryAfter = next
case 408, 500, 502, 503, 504:
auth.StatusMessage = "transient upstream error"

View File

@@ -5,6 +5,7 @@ import (
"strconv"
"strings"
"sync"
"sync/atomic"
"time"
baseauth "github.com/router-for-me/CLIProxyAPI/v6/internal/auth"
@@ -14,6 +15,8 @@ import (
type Auth struct {
// ID uniquely identifies the auth record across restarts.
ID string `json:"id"`
// Index is a monotonically increasing runtime identifier used for diagnostics.
Index uint64 `json:"-"`
// Provider is the upstream provider key (e.g. "gemini", "claude").
Provider string `json:"provider"`
// FileName stores the relative or absolute path of the backing auth file.
@@ -55,6 +58,8 @@ type Auth struct {
// Runtime carries non-serialisable data used during execution (in-memory only).
Runtime any `json:"-"`
indexAssigned bool `json:"-"`
}
// QuotaState contains limiter tracking data for a credential.
@@ -87,6 +92,12 @@ type ModelState struct {
UpdatedAt time.Time `json:"updated_at"`
}
var authIndexCounter atomic.Uint64
func nextAuthIndex() uint64 {
return authIndexCounter.Add(1) - 1
}
// Clone shallow copies the Auth structure, duplicating maps to avoid accidental mutation.
func (a *Auth) Clone() *Auth {
if a == nil {
@@ -115,6 +126,20 @@ func (a *Auth) Clone() *Auth {
return &copyAuth
}
// EnsureIndex returns the global index, assigning one if it was not set yet.
func (a *Auth) EnsureIndex() uint64 {
if a == nil {
return 0
}
if a.indexAssigned {
return a.Index
}
idx := nextAuthIndex()
a.Index = idx
a.indexAssigned = true
return idx
}
// Clone duplicates a model state including nested error details.
func (m *ModelState) Clone() *ModelState {
if m == nil {
@@ -152,11 +177,29 @@ func (a *Auth) AccountInfo() (string, string) {
}
}
}
if a.Metadata != nil {
if v, ok := a.Metadata["email"].(string); ok {
return "oauth", v
// For iFlow provider, prioritize OAuth type if email is present
if strings.ToLower(a.Provider) == "iflow" {
if a.Metadata != nil {
if email, ok := a.Metadata["email"].(string); ok {
email = strings.TrimSpace(email)
if email != "" {
return "oauth", email
}
}
}
}
// Check metadata for email first (OAuth-style auth)
if a.Metadata != nil {
if v, ok := a.Metadata["email"].(string); ok {
email := strings.TrimSpace(v)
if email != "" {
return "oauth", email
}
}
}
// Fall back to API key (API-key auth)
if a.Attributes != nil {
if v := a.Attributes["api_key"]; v != "" {
return "api_key", v
@@ -259,6 +302,7 @@ func parseTimeValue(v any) (time.Time, bool) {
time.RFC3339,
time.RFC3339Nano,
"2006-01-02 15:04:05",
"2006-01-02 15:04",
"2006-01-02T15:04:05Z07:00",
}
for _, layout := range layouts {

View File

@@ -333,6 +333,8 @@ func (s *Service) ensureExecutorsForAuth(a *coreauth.Auth) {
s.coreManager.RegisterExecutor(executor.NewAIStudioExecutor(s.cfg, a.ID, s.wsGateway))
}
return
case "antigravity":
s.coreManager.RegisterExecutor(executor.NewAntigravityExecutor(s.cfg))
case "claude":
s.coreManager.RegisterExecutor(executor.NewClaudeExecutor(s.cfg))
case "codex":
@@ -629,11 +631,15 @@ func (s *Service) registerModelsForAuth(a *coreauth.Auth) {
models = registry.GetGeminiModels()
case "vertex":
// Vertex AI Gemini supports the same model identifiers as Gemini.
models = registry.GetGeminiModels()
models = registry.GetGeminiVertexModels()
case "gemini-cli":
models = registry.GetGeminiCLIModels()
case "aistudio":
models = registry.GetAIStudioModels()
case "antigravity":
ctx, cancel := context.WithTimeout(context.Background(), 15*time.Second)
models = executor.FetchAntigravityModels(ctx, a, s.cfg)
cancel()
case "claude":
models = registry.GetClaudeModels()
if entry := s.resolveConfigClaudeKey(a); entry != nil && len(entry.Models) > 0 {

View File

@@ -14,6 +14,7 @@ type Record struct {
Model string
APIKey string
AuthID string
AuthIndex uint64
Source string
RequestedAt time.Time
Failed bool

View File

@@ -8,4 +8,5 @@ const (
FormatGemini Format = "gemini"
FormatGeminiCLI Format = "gemini-cli"
FormatCodex Format = "codex"
FormatAntigravity Format = "antigravity"
)