CLIProxyAPI

Merge pull request #1699 from 123hi123/fix/antigravity-primary-model-fallback

fix(antigravity): keep primary model list and backfill empty auths

Luis Pater · 2026-02-26 04:28:29 +08:00

24bcfd9c03

Merge pull request #1682 from sususu98/fix/tool-result-image-parts

fix(antigravity): place tool_result images in functionResponse.parts and unify mimeType

Luis Pater · 2026-02-25 23:14:35 +08:00

816fb4c5da

Merge pull request #1664 from ciberponk/pr/responses-compaction-compat

feat: add codex responses compatibility for compaction payloads

Luis Pater · 2026-02-25 01:21:59 +08:00

v6.8.28 d24ea4ce2a

Merge pull request #1687 from lyd123qw2008/fix/codex-refresh-token-reused-no-retry

fix(codex): stop retrying refresh_token_reused errors

Luis Pater · 2026-02-25 01:19:30 +08:00

2c30c981ae

Merge pull request #1685 from lyd123qw2008/fix/auth-auto-refresh-interval

fix(auth): respect configured auto-refresh interval

Luis Pater · 2026-02-25 01:13:47 +08:00

aa1da8a858

Merge pull request #1676 from piexian/feat/qwen-quota-handling-clean

feat(qwen): add rate limiting and quota error handling

Luis Pater · 2026-02-25 01:07:55 +08:00

f1e9a787d7

Merge pull request #1683 from dusty-du/codex/device-login-flow

Add additive Codex device-code login flow

Luis Pater · 2026-02-25 00:50:48 +08:00

v6.8.27 c66cb0afd2

Merge pull request #1680 from canxin121/fix/responses-stream-error-chunks

fix(responses): emit schema-valid SSE chunks

Luis Pater · 2026-02-25 00:49:06 +08:00

fb48eee973

Merge pull request #1701 from router-for-me/openai

Revert "Merge pull request #1627 from thebtf/fix/reasoning-effort-clamping"

Luis Pater · 2026-02-25 00:46:13 +08:00

bb44e5ec44

fix(antigravity): deep copy cached model metadata

comalot · 2026-02-24 20:14:01 +08:00

514ae341c8

Revert "Merge pull request #1627 from thebtf/fix/reasoning-effort-clamping"

hkfires · 2026-02-24 19:47:53 +08:00

0659ffab75

fix(antigravity): keep primary model list and backfill empty auths

comalot · 2026-02-24 16:16:44 +08:00

8ce07f38dd

Merge pull request #1663 from rensumo/main

feat: implement credential-based round-robin for gemini-cli

Luis Pater · 2026-02-24 06:02:50 +08:00

7cb398d167

Merge pull request #1654 from alexey-yanchenko/feature/pass-file-inputs

Pass file input from /chat/completions and /responses to codex and claude

Luis Pater · 2026-02-24 05:53:11 +08:00

c3e12c5e58

Merge pull request #1643 from alexey-yanchenko/fix/gemini-prompt-tokens

Fix usage convertation from gemini response to openai format

Luis Pater · 2026-02-24 05:46:13 +08:00

1825fc7503

Merge pull request #1527 from HEUDavid/feat/auth-hook

feat(auth): add post-auth hook mechanism

Luis Pater · 2026-02-24 05:33:13 +08:00

48732ba05e

fix(responses): reject invalid SSE data JSON

Guard the openai-response streaming path against truncated/invalid SSE data payloads by validating data: JSON before forwarding; surface a 502 terminal error instead of letting clients crash with JSON parse errors.

canxin121 · 2026-02-24 01:42:54 +08:00

acf483c9e6

test(codex): log non-retryable refresh error and cover single-attempt behavior

lyd123qw2008 · 2026-02-23 22:41:33 +08:00

3b3e0d1141

fix(codex): stop retrying refresh_token_reused errors

lyd123qw2008 · 2026-02-23 22:31:30 +08:00

7acd428507

fix(auth): respect configured auto-refresh interval

lyd123qw2008 · 2026-02-23 22:07:50 +08:00

450d1227bd

Add additive Codex device-code login flow

test · 2026-02-23 06:30:04 -05:00

492b9c46f0

fix(antigravity): place tool_result images in functionResponse.parts and unify mimeType

Move base64 image data from Claude tool_result into functionResponse.parts
as inlineData instead of outer sibling parts, preventing context bloat.
Unify all inlineData field naming to camelCase mimeType across Claude,
OpenAI, and Gemini translators. Add comprehensive edge case tests and
Gemini-side regression test for functionResponse.parts preservation.

sususu98 · 2026-02-23 13:38:21 +08:00

4e26182d14

revert: translator changes (path guard)

CI blocks PRs that modify internal/translator. Revert translator edits and keep only the /v1/responses streaming error-chunk fix; file an issue for translator conformance work.

canxin121 · 2026-02-23 13:30:43 +08:00

eb7571936c

fix(responses): include model and usage in translated streams

Ensure response.created and response.completed chunks produced by the OpenAI/Gemini/Claude translators always include required fields (response.model and response.usage) so clients validating Responses SSE do not fail schema validation.

canxin121 · 2026-02-23 13:22:06 +08:00

5382764d8a

fix(openai): emit valid responses stream error chunks

When /v1/responses streaming fails after headers are sent, we now emit a type=error chunk instead of an HTTP-style {error:{...}} payload, preventing AI SDK chunk validation errors.

canxin121 · 2026-02-23 12:59:50 +08:00

49c8ec69d0

feat(qwen): add rate limiting and quota error handling

- Add 60 requests/minute rate limiting per credential using sliding window
- Detect insufficient_quota errors and set cooldown until next day (Beijing time)
- Map quota errors (HTTP 403/429) to 429 with retryAfter for conductor integration
- Cache Beijing timezone at package level to avoid repeated syscalls
- Add redactAuthID function to protect credentials in logs
- Extract wrapQwenError helper to consolidate error handling

piexian · 2026-02-23 00:38:46 +08:00

3b421c8181

Fixed: #1675

fix(gemini): add model definitions for Gemini 3.1 Pro High and Image

Luis Pater · 2026-02-23 00:12:57 +08:00

v6.8.26 713388dd7b

Merge pull request #1522 from soilSpoon/feature/canceled

feature(proxy): Adds special handling for client cancellations in proxy error handler

Luis Pater · 2026-02-22 22:02:59 +08:00

e6c7af0fa9

fix(gemini): update min Thinking value and add Gemini 3.1 Pro Preview model definition

Luis Pater · 2026-02-22 21:51:32 +08:00

v6.8.25 d210be06c2

refactor: simplify context_management compatibility handling

fan · 2026-02-21 22:20:48 +08:00

afc8a0f9be

Merge pull request #1662 from matchch/contribute/cache-user-id

feat: add cache-user-id toggle for Claude cloaking

Luis Pater · 2026-02-21 20:51:30 +08:00

v6.8.24 d6ec33e8e1

fix(gemini): correct Created timestamps for Gemini 3.1 Pro Preview model definitions

Luis Pater · 2026-02-21 20:47:47 +08:00

081cfe806e

feat(gemini): add Gemini 3.1 Pro Preview model definitions

hkfires · 2026-02-21 20:42:29 +08:00

c1c62a6c04

feat: support responses compaction payload compatibility for codex translator

ciberponk · 2026-02-21 12:56:10 +08:00

d693d7993b

feat: implement credential-based round-robin for gemini-cli virtual auths

Changes the RoundRobinSelector to use two-level round-robin when
gemini-cli virtual auths are detected (via gemini_virtual_parent attr):
- Level 1: cycle across credential groups (parent accounts)
- Level 2: cycle within each group's project auths

Credentials start from a random offset (rand.IntN) for fair distribution.
Non-virtual auths and single-credential scenarios fall back to flat RR.

Adds 3 test cases covering multi-credential grouping, single-parent
fallback, and mixed virtual/non-virtual fallback.

rensumo · 2026-02-21 12:49:48 +08:00

5936f9895c

feat: add cache-user-id toggle for Claude cloaking

Default to generating a fresh random user_id per request instead of
reusing cached IDs. Add cache-user-id config option to opt in to the
previous caching behavior.

- Add CacheUserID field to CloakConfig
- Extract user_id cache logic to dedicated file
- Generate fresh user_id by default, cache only when enabled
- Add tests for both paths

matchch · 2026-02-21 12:31:20 +08:00

2fdf5d2793

Merge pull request #1660 from Grivn/fix/claude-token-url

fix(claude): use api.anthropic.com for OAuth token exchange

Luis Pater · 2026-02-20 21:52:08 +08:00

v6.8.23 7b0eb41ebc

fix(claude): use api.anthropic.com for OAuth token exchange

console.anthropic.com is now protected by a Cloudflare managed challenge
that blocks all non-browser POST requests to /v1/oauth/token, causing
`-claude-login` to fail with a 403 error.

Switch to api.anthropic.com which hosts the same OAuth token endpoint
without the Cloudflare managed challenge.

Fixes #1659

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Grivn · 2026-02-20 20:11:27 +08:00

ef5901c81b

Merge pull request #1652 from thebtf/fix/claude-translator-arguments

fix(translator): handle tool call arguments in codex→claude streaming translator

Luis Pater · 2026-02-20 19:50:20 +08:00

d4829c82f7

Merge pull request #1644 from possible055/main

feat: add Gemini 3.1 Pro Preview model definition

Luis Pater · 2026-02-20 19:44:59 +08:00

a5f4166a9b

Pass file input from /chat/completions and /responses to codex and claude

Alexey Yanchenko · 2026-02-20 10:25:44 +07:00

0cbfe7f457

fix: prevent duplicate function call arguments when delta events precede done

Non-spark codex models (gpt-5.3-codex, gpt-5.2-codex) stream function call
arguments via multiple delta events followed by a done event. The done handler
unconditionally emitted the full arguments, duplicating what deltas already
streamed. This produced invalid double JSON that Claude Code couldn't parse,
causing tool calls to fail with missing parameters and infinite retry loops.

Add HasReceivedArgumentsDelta flag to track whether delta events were received.
The done handler now only emits arguments when no deltas preceded it (spark
models), while delta-based streaming continues to work for non-spark models.

Kirill Turanskiy · 2026-02-19 23:18:14 +03:00

1cc21cc45b

fix: handle response.function_call_arguments.done in codex→claude streaming translator

Some Codex models (e.g. gpt-5.3-codex-spark) send function call arguments
in a single "done" event without preceding "delta" events. The streaming
translator only handled "delta" events, causing tool call arguments to be
lost — resulting in empty tool inputs and infinite retry loops in clients
like Claude Code.

Emit the full arguments from the "done" event as a single input_json_delta
so downstream clients receive the complete tool input.

Kirill Turanskiy · 2026-02-19 23:18:14 +03:00

07cf616e2b

test(handlers): add tests for passthrough headers behavior in WriteErrorResponse

Luis Pater · 2026-02-19 21:49:44 +08:00

v6.8.22 4445a165e9

Merge branch 'codex/pr-1626' into dev

Luis Pater · 2026-02-19 21:33:23 +08:00

e92e2af71a

feat: add passthrough headers configuration

- Introduced `passthrough-headers` option in configuration to control forwarding of upstream response headers.
- Updated handlers to respect the passthrough headers setting.
- Added tests to verify behavior when passthrough is enabled or disabled.

Luis Pater · 2026-02-19 21:31:29 +08:00

a6bdd9a652

Merge pull request #1645 from tinyc0der/fix/antigravity-tool-result-json

fix(antigravity): prevent invalid JSON when tool_result has no content

Luis Pater · 2026-02-19 21:01:25 +08:00

349a6349b3

fix(antigravity): prevent invalid JSON when tool_result has no content

sjson.SetRaw with an empty string produces malformed JSON (e.g. "result":}).
This happens when a Claude tool_result block has no content field, causing
functionResponseResult.Raw to be "". Guard against this by falling back to
sjson.Set with an empty string only when .Raw is empty.

TinyCoder · 2026-02-19 17:08:39 +07:00

00822770ec

feat: add Gemini 3.1 Pro Preview model definition

apparition · 2026-02-19 17:43:08 +08:00

1a0ceda0fc

Fix usage convertation from gemini response to openai format

Alexey Yanchenko · 2026-02-19 15:34:59 +07:00

b9ae4ab803

1772 Commits