Commit Graph

3123 Commits

Author SHA1 Message Date
diegosouzapw
775c87bb8f docs(changelog): document #2321 #2341 #2346 #2348 #2352 fixes 2026-05-18 09:19:18 -03:00
diegosouzapw
33928afec0 fix(ui/tooltip): render in portal + clamp to viewport (#2352)
When the shared <Tooltip> appeared inside a modal (combo editor) or any
ancestor with overflow:hidden/auto, long labels were clipped — the
absolute-positioned <span> stayed inside the modal's stacking context.

Render the tooltip via createPortal to document.body by default
(usePortal prop, defaults to true). Coordinates are computed from the
trigger's getBoundingClientRect on each show and written directly to
the tooltip ref's .style — that avoids the cascading-render warning
that setState-inside-effect triggers, while still doing one synchronous
measure-and-position pass before the user sees any flicker.

Coordinates are clamped to the viewport bounds so a trigger near the
right edge produces a tooltip that stays on-screen instead of bleeding
off. Adds an optional multiline prop that swaps the legacy
whitespace-nowrap clamp for max-w-xs whitespace-normal break-words for
explanation strings (combo strategy help, etc.).

Backward compat: existing call sites do not need to change; the portal
+ clamp behavior is transparent to consumers.
2026-05-18 09:18:25 -03:00
diegosouzapw
9c5b75e3bf test(codex): regression guard for effort suffix priority (#2331)
Adds a deterministic test for the rawEffort resolution chain in
open-sse/executors/codex.ts so a future refactor that flips the priority
back (the bug we just fixed) trips an explicit assertion. Also verifies
end-to-end behavior of the priority order: modelEffort > explicitReasoning
> requestReasoningEffort > fallbackReasoningEffort.

The source fix itself landed earlier on this branch; this commit only
adds the test coverage.
2026-05-18 09:12:36 -03:00
diegosouzapw
d62b128144 fix(account-fallback): classify Anthropic 'Usage Limit Reached' as quota (#2321)
Claude Code Pro/Team users hit a 429 cascade where every Claude account
got marked rate-limited for only a few seconds, retried, hit 429 again,
and the cycle exhausted all accounts until the 5h subscription window
finally reset. Root cause: Anthropic OAuth 429 bodies carry phrases like
'Usage Limit Reached' or 'Claude Pro usage limit reached' that don't
contain the word 'quota'. classifyErrorText fell through to
RATE_LIMIT_EXCEEDED with the OAuth ~5s base cooldown.

Three targeted changes in open-sse/services/accountFallback.ts:

1. classifyErrorText: recognize the Anthropic OAuth phrases ('usage limit
   reached', 'claude pro usage limit', 'you've reached your usage limit'
   and variants) and return QUOTA_EXHAUSTED.

2. checkFallbackError: new branch that, when the classifier flags
   QUOTA_EXHAUSTED for a non-credits/non-daily case, applies a deliberate
   1h cooldown (subscription quotas need real time to reset; the existing
   COOLDOWN_MS.paymentRequired is only 2 minutes).

3. parseRetryFromErrorText: parse absolute ISO 8601 timestamps in the
   body ('Try again at 2026-05-17T10:00:00Z', 'Please wait until ...').
   When present, honor the upstream's stated recovery time instead of
   the 1h default.

Cross-provider fallback for direct (non-combo) calls remains a separate
feature — operators wanting that should route Claude through a combo
that includes openrouter/claude or another fallback target.
2026-05-18 09:11:32 -03:00
diegosouzapw
fe3ab7a836 fix(combo/validator): accept reasoning_content as valid output (#2341)
`validateResponseQuality` flagged any response with `content: null` as
empty and triggered a false-positive 502 combo fallback, even when the
upstream returned the entire answer in `reasoning_content`. This
affected every reasoning model that omits `content` by design:
moonshotai/Kimi-K2.5-TEE, zai-org/GLM-5-TEE, zai-org/GLM-4.7-TEE,
DeepSeek-R1 and Qwen-thinking variants via Chutes.ai, Nvidia, and other
OpenAI-compatible gateways.

Treat a non-empty `reasoning_content` (or its legacy `reasoning` alias)
as valid content. Empty/whitespace-only strings still fall through to
the existing empty-content rejection so we don't weaken the guard.

Backward compat verified: regular content-only and tool_calls-only
responses still validate without change.
2026-05-18 09:10:35 -03:00
diegosouzapw
2af324f6ee fix(docker): ship Dashboard Docs markdown in the container image (#2348)
`.dockerignore` excluded everything under `docs/` except `openapi.yaml`,
which broke the in-product Docs viewer at `/docs/*` with "ENOENT: no
such file or directory, open '/app/docs/SETUP_GUIDE.md'" for every help
page. The previous block-list was meant to keep the image small but it
hid the ~5 MB English markdown tree that the viewer actually reads.

Replace the block with a targeted exclude of the heavy assets that
account for ~45 MB of the original ~50 MB:
- docs/i18n/** (translated copies — viewer falls back to English)
- docs/screenshots/**
- docs/diagrams/exported/** plus raster/SVG variants

Note: Go filepath.Match (Docker's matcher) treats `*` as not crossing
`/`, so the existing `*.md` rule still excludes only root-level
markdown — nested `docs/**/*.md` is implicitly kept without a
re-include rule that would have brought i18n back.

Added a regression test that parses .dockerignore and asserts the
critical English docs survive the filter while the heavy paths still
do not.
2026-05-18 09:09:45 -03:00
diegosouzapw
5da9b1b778 fix(auto-routing): replace bare getSettings() with getCachedSettings (#2346)
The auto-routing path in src/sse/handlers/chat.ts called `getSettings()`
on every `auto` / `auto/\*` request, but the symbol was never imported in
this module (only `getCachedSettings` was). The runtime ReferenceError
crashed the request with a 500 before any routing could happen.

Replace the call with `getCachedSettings` (already imported, identical
return shape, plus the cached variant benefits the auto-routing hot path).
A regression-guard test scans chat.ts to fail loudly if a future refactor
re-introduces the unimported call pattern.
2026-05-18 09:08:52 -03:00
diegosouzapw
b464946b1b docs(changelog): credit contributors for PRs 2362, 2364, 2366, 2369 2026-05-18 08:38:21 -03:00
Paijo
c4fe3d63be feat(content): extend providers with video, audio, TTS, music capabilities (#2369)
Integrated into release/v3.8.0
2026-05-18 08:37:44 -03:00
Paijo
dd24571949 feat(providers): add Veo AI Free as web wrapper provider (#2366)
Integrated into release/v3.8.0
2026-05-18 08:35:34 -03:00
Paijo
41b7f13b27 feat(providers): add Replicate as free provider (#2364)
Integrated into release/v3.8.0
2026-05-18 08:34:00 -03:00
terence71-glitch
6c488d6d2a fix: avoid redundant Claude Code message clone (#2362)
Integrated into release/v3.8.0
2026-05-18 08:30:47 -03:00
diegosouzapw
2bbd0fff45 Merge feat/gemini-auth-ui (PR3 Gemini) into release/v3.8.0
# Conflicts:
#	src/app/(dashboard)/dashboard/providers/[id]/page.tsx
#	src/i18n/messages/en.json
2026-05-18 03:21:56 -03:00
diegosouzapw
1dae73cfef Merge feat/gemini-auth-api-routes (PR2 Gemini) into release/v3.8.0
# Conflicts:
#	src/shared/validation/schemas.ts
2026-05-18 03:01:51 -03:00
diegosouzapw
f67c7d9383 Merge feat/gemini-auth-libs (PR1 Gemini) into release/v3.8.0 2026-05-18 02:58:51 -03:00
diegosouzapw
0b2085db83 Merge feat/claude-auth-ui (PR3 Claude) into release/v3.8.0 2026-05-18 02:58:42 -03:00
diegosouzapw
b9126e51f7 Merge feat/claude-auth-api-routes (PR2 Claude) into release/v3.8.0 2026-05-18 02:58:36 -03:00
diegosouzapw
1e709fdef1 Merge feat/claude-auth-libs (PR1 Claude) into release/v3.8.0 2026-05-18 02:58:27 -03:00
diegosouzapw
855dc86ea8 fix(dashboard): add missing comma in Gemini i18n JSON (PR3) 2026-05-18 02:45:51 -03:00
diegosouzapw
b3cb5b68d8 feat(dashboard): add Gemini CLI auth import/export UI + i18n (PR3) 2026-05-18 02:43:15 -03:00
diegosouzapw
75c1d2ead2 feat(dashboard): add Claude Code auth import/export UI + i18n (PR3) 2026-05-18 02:26:08 -03:00
diegosouzapw
877aafdb99 feat(api): add Gemini CLI auth import/export API routes + schemas (PR2) 2026-05-18 01:46:07 -03:00
diegosouzapw
3786e929d6 feat(api): add Claude Code auth import/export API routes + schemas (PR2) 2026-05-18 01:45:54 -03:00
diegosouzapw
c6a51605ce feat(oauth): add Gemini CLI auth import/export libs + CLI_TOOLS registration (PR1) 2026-05-18 01:12:44 -03:00
diegosouzapw
09fb5cdf34 feat(oauth): add Claude Code auth import/export libs + tests (PR1) 2026-05-18 01:07:38 -03:00
diegosouzapw
4eb5ba5fb8 docs(changelog): add entry for PR #2355 2026-05-18 00:25:07 -03:00
Raxxoor
f6364427ce fix: emit stream errors in client protocol (#2355)
Integrated into release/v3.8.0
2026-05-18 00:24:38 -03:00
diegosouzapw
7a40e0f547 docs(changelog): add entry for PR #2354 2026-05-17 22:59:23 -03:00
Cong Vu Chi
6d4ee4f85a fix: allow bracketed combo names (#2354)
Integrated into release/v3.8.0
2026-05-17 22:58:42 -03:00
diegosouzapw
749b945fdf docs(changelog): add entries for PRs #2351 #2350
- feat(claude-code): semantic passthrough (#2351 — thanks @terence71-glitch)
- fix(usage): flat cached_tokens/reasoning_tokens extraction (#2350 — thanks @TF0rd)
2026-05-17 21:12:11 -03:00
Tyler Ford
d6d585e200 fix: extract flat cached_tokens/reasoning_tokens from OpenAI-compatible usage objects (#2350)
Integrated into release/v3.8.0 — flat cached_tokens/reasoning_tokens extraction for Xiaomi MiMo and similar providers
2026-05-17 21:11:06 -03:00
terence71-glitch
db8a2dc735 fix: preserve Claude Code messages on direct routes (#2351)
Integrated into release/v3.8.0 — Claude Code semantic passthrough for /v1/messages payloads
2026-05-17 21:09:02 -03:00
diegosouzapw
9b8dc4d6db docs(changelog): add entries for PRs #2326 #2327 #2335 #2349 #2344 #2339 #2322 #2329 #2338 #2340 2026-05-17 20:15:02 -03:00
Paijo
1be18f5530 feat(providers): add Microsoft Copilot Web wrapper (#2340)
Integrated into release/v3.8.0 — fixed session pool security issue (per-token isolation) and added unit tests
2026-05-17 20:02:02 -03:00
Paijo
193e7a6417 feat(routing): LGKP remembers last good account per provider (#2338)
Integrated into release/v3.8.0 — added unit tests for connectionId-aware getLKGP/setLKGP
2026-05-17 19:54:32 -03:00
Michael
b191173ae1 Fix Providers empty state blocking first provider setup 2026-05-17 19:47:42 -03:00
backryun
1f27c344d4 chore: improve huggingface provider support (#2322)
Integrated into release/v3.8.0 — migrates HuggingFace to router.huggingface.co/v1 endpoint, adds dynamic model discovery, refreshes ASR/TTS catalog
2026-05-17 19:44:53 -03:00
Paijo
aefd9bbbc7 feat(providers): add Hackclub AI as free provider (#2339)
Integrated into release/v3.8.0 — adds Hackclub AI as free aggregator with 30+ models, auth optional
2026-05-17 19:43:58 -03:00
Paijo
c2451038a0 feat(providers): add GitHub Models as free provider (#2344)
Integrated into release/v3.8.0 — adds GitHub Models as a free provider with 14 models (GPT-4.1, o3, DeepSeek-R1, Llama 4, Grok 3 etc.)
2026-05-17 19:40:50 -03:00
Hernan Javier Ardila Sanchez
749197466f fix(translator): use reasoning cache before empty string fallback for DeepSeek tool_calls (#2349)
Integrated into release/v3.8.0 — fixes DeepSeek 400 errors by using cached reasoning_content before falling back to empty string for tool_call messages
2026-05-17 19:40:10 -03:00
terence71-glitch
0e2d04526c fix: honor Codex reasoning suffix aliases (#2335)
Integrated into release/v3.8.0 — fixes issue #2331: Codex model suffix aliases now correctly override client-injected reasoning defaults
2026-05-17 19:38:25 -03:00
thepigdestroyer
696e16b361 fix(claude): cap-aware thinking budget fit for max_tokens (#2327)
Integrated into release/v3.8.0 — fixes HTTP 400 max_tokens > 128000 from Anthropic when using high-effort thinking with Opus 4.7
2026-05-17 19:37:40 -03:00
thepigdestroyer
0f15797e01 fix(v1/messages): default to non-stream for Claude format when ambiguous (#2326)
Integrated into release/v3.8.0 — fixes STREAM_EARLY_EOF on POST /v1/messages when stream is omitted
2026-05-17 19:36:57 -03:00
Diego Rodrigues de Sa e Souza
891a9e4125 Merge pull request #2337 from diegosouzapw/worktree-fix+providers-missing-button-i18n
fix(providers): fix missing i18n keys and add 'Add Provider' button to empty state
2026-05-17 17:51:43 -03:00
Diego Rodrigues de Sa e Souza
9c69a20ea5 Merge pull request #2343 from diegosouzapw/feat/codex-auth-import-bulk
feat(codex): bulk import Codex auth.json — multi-file, paste, ZIP
2026-05-17 17:29:49 -03:00
diegosouzapw
3748b0238d chore(merge): resolve conflicts for bulk import — keep both schemas and all i18n keys 2026-05-17 17:28:41 -03:00
Diego Rodrigues de Sa e Souza
6e6cb5a0b5 Merge pull request #2336 from diegosouzapw/feat/codex-auth-import-single
feat(codex): import single Codex auth.json as OAuth connection
2026-05-17 17:14:52 -03:00
diegosouzapw
3227c9ffe3 chore(merge): resolve page.tsx conflict — keep ImportCodexAuthModal and ApplyCodexAuthModal 2026-05-17 17:13:50 -03:00
Diego Rodrigues de Sa e Souza
cf0006fd7b Merge pull request #2332 from diegosouzapw/feat/codex-auth-apply-modal
feat(codex-auth): rename export + gate Apply Local behind confirmation modal
2026-05-17 17:08:11 -03:00
diegosouzapw
25f3fe1ac5 feat(codex): bulk import Codex auth.json — multi-file, paste, ZIP
Adds three input modes for importing multiple Codex accounts at once,
all feeding a single partial-failure backend endpoint.

- `codexAuthZipExtract.ts`: safe ZIP extraction via fflate — rejects
  path traversal (../ and absolute paths), per-file 256 KB cap, 10 MB
  total cap, max 50 .json entries
- `POST /api/providers/codex-auth/zip-extract`: server-side ZIP
  extraction returning [{name, json, parseError}] to the client
- `POST /api/providers/codex-auth/import-bulk`: iterates entries,
  partial-failure semantics (always 200), per-entry audit log
  `provider.credentials.imported` + summary `bulk_imported`
- `importCodexAuthBulkSchema`: max 50 entries, email validation
- `<ImportCodexAuthModal>`: Single/Bulk top tabs; Bulk has Upload
  files, Paste list (JSON array or --- separator), ZIP sub-modes;
  live entry preview list; overwrite checkbox; result panel
- Install `fflate@0.8.3` (pure TypeScript, zero native deps)
- 29 unit tests: 12 ZIP safety cases + 17 schema/parser/shape cases
2026-05-17 16:52:43 -03:00