mirror of
https://github.com/diegosouzapw/OmniRoute.git
synced 2026-08-18 05:02:15 +03:00
6ff2e7b2c2fdc19ed2e9637ea4325af689728ea3
8 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
Diego Rodrigues de Sa e Souza
|
f68856695d |
maint: follow-up cherry-pick fix-in-place #9693 (conflict-resolved fallback) (#9887)
* fix(web-tools): anchor tool contract at prompt tail + user-turn reminder The <tool> contract from prepareToolMessages was prepended as the first system message. Web executors fold all system messages into one block, so with agentic clients whose system prompts exceed ~28K chars the contract sat at the head of a huge block and web models ignored it, refusing tool calls with "tool X is not in my tool set" (chatgpt-web, 0/3 at 30K chars). Two changes, both required in testing: - Dual placement: the full contract now rides as a trailing system message (folds to the tail of the system block) and a one-line reminder naming the tools is appended to the latest user message. - Rewording: the contract now frames injected tools as client tools invoked via a plain-text protocol, distinct from the model's native tool registry (web.run, python.exec, ...), and instructs the model to never claim they are unavailable. Without this the model resolved tool names against its native registry and refused even when it had seen the contract. Measured on cgpt-web gpt-5.5-thinking/gpt-5.6-thinking/o3: prepend 0/3 tool calls at 30K chars; dual placement 16/17 across 30K-250K system prompts, 30-tool sets, multi-turn tool history, streaming, and 3-way concurrency, with no spurious calls on no-tool prompts. Known limit: ~40K-char single user messages still flake (2/3) due to the upstream model's own injection heuristics. All prepareToolMessages consumers parse system messages position-independently and select the current user turn by role scan, so the trailing system message is shape-safe for every web executor. * test(web-tools): cover contract placement edge cases --------- Co-authored-by: Ryan Brosas <ryanjoserbrosas@gmail.com> |
||
|
Diego Rodrigues de Sa e Souza
|
babc75f058 |
feat(chatgpt-web): harden prompt-emulated tool contract for thinking models (#7679) (#9422)
Merge-train validated (tip 6ce4effef8). Vitest failures confirmed as base-red (#9679). |
||
|
Diego Rodrigues de Sa e Souza
|
d969555417 | fix(security): require explicit tool envelope to prevent bare JSON tool_calls (#9343) | ||
|
Diego Rodrigues de Sa e Souza
|
19d91d82e2 |
Release v3.8.34 (#4614)
* chore(release): open v3.8.34 development cycle * chore(quality): release-green pre-flight validator + nightly signal (C+D) (#4622) C — scripts/quality/validate-release-green.mjs (npm run check:release-green): reproduces the release-equivalent validation (typecheck, eslint, db-rules, public-creds, full unit, vitest, ratchets, optional --with-build package-artifact) against the current working tree and classifies each red as HARD (real defect, exit 1) vs DRIFT (ratchet — reported, never affects exit / never blocks). Pure helpers exported + orchestration behind a direct-run guard; unit-tested. D — .github/workflows/nightly-release-green.yml: runs C on the active release branch nightly (and on workflow_dispatch) and opens/updates a single tracking issue on HARD failures. Never a required check, never touches a contributor PR. Closes the gap where the full gate (ci.yml) only ran on the release PR, so reds accrued silently on release/** and surfaced in 40-min layers at release time. Non-blocking by construction; drift is the maintainer's to rebaseline at release. Co-authored-by: Diego Rodrigues de Sa e Souza <diego.souza@cdwasolutions.com.br> * fix(providers): show revealed connection API keys (#4583) Integrated into release/v3.8.34 * fix(resilience): respect upstream retry hint toggle (#4585) Integrated into release/v3.8.34 * feat(settings): expose stream recovery feature flags (#4586) Integrated into release/v3.8.34 * fix(logs): make active request stale sweep configurable (#4599) Integrated into release/v3.8.34 * fix(plugin): auto-prefix providerId with 'opencode-' for OC 1.17.8+ native gate (#4527) Integrated into release/v3.8.34 (supersedes #4445) * fix(models): treat unknown output caps as unset (#4584) Integrated into release/v3.8.34 * fix(executors): strip temperature for GitHub Copilot gpt-5.4 family (#4564) Integrated into release/v3.8.34 (rebuilt onto tip) * fix(oauth): update Qwen OAuth URLs from chat.qwen.ai to qwen.ai (#4561) Integrated into release/v3.8.34 (rebuilt onto tip) * fix(api/settings): prevent cached /api/settings responses (port from 9router#951) (#4566) Integrated into release/v3.8.34 (rebuilt onto tip) * feat(audio): MiniMax T2A v2 TTS dispatch in audioSpeech (port #1043) (#4553) Integrated into release/v3.8.34 (rebuilt onto tip) * fix(dashboard): surface manual config CTA when Open Claw CLI auto-detect fails (#4562) Integrated into release/v3.8.34 (rebuilt onto tip) * feat(providers): optional model ID for custom API-key validation (#4555) Integrated into release/v3.8.34 (rebuilt onto tip) * fix(cli): align data dir and env loading with runtime (#4607) Integrated into release/v3.8.34 (rebuilt onto tip) * fix(quota): expose Bailian quota windows (#4610) Integrated into release/v3.8.34 (rebuilt onto tip) * fix: retain provider cooldowns for configured max window (#4588) Integrated into release/v3.8.34 (rebuilt — bundled commits stripped) * fix: reject invalid provider cooldown bounds (#4589) Integrated into release/v3.8.34 (rebuilt — bundled commits stripped) * fix: preserve production combo metrics on shadow eviction (#4590) Integrated into release/v3.8.34 (rebuilt — bundled commits stripped) * fix(stream): estimate input tokens when upstream reports prompt_tokens=0 (#4615) Integrated into release/v3.8.34 (rebuilt onto tip) * fix(catalog): shorten no-thinking gateway prefix to no-think/ (#4525) Integrated into release/v3.8.34 (rebuilt — kept only the prefix rename, dropped stale-base reverts) * fix(relay): apply IP rate limit to bifrost sidecar (#4593) Integrated into release/v3.8.34 (rebuilt onto tip; merge before #4612) * fix(bifrost): finalize SSE relay usage after stream (#4612) Integrated into release/v3.8.34 (rebuilt + reconciled with #4593) * feat(compression): per-request `x-omniroute-compression` header (Phase 3) (#4645) * docs(compression): Phase 3 per-request header design spec Approved brainstorming output for the x-omniroute-compression header: header-first precedence, name-first combo matching (Decision A), explicit value bypasses auto-trigger (Decision B), DerivedPlan.source, and the X-OmniRoute-Compression response header. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * docs(compression): Phase 3 per-request header implementation plan 4-task TDD plan (resolver header-first + source, parser, chatCore wiring + response header, docs/file-size) with full code and exact commands. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(compression): header-first resolver + plan source (Phase 3 core) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(compression): resolveCompressionHeader parser (Phase 3) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(compression): wire x-omniroute-compression header + response header (Phase 3) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * refactor(compression): extract plan-resolution leaf (planResolution.ts) under size cap (Phase 3) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * docs(compression): document x-omniroute-compression header (Phase 3) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(compression): harden named-combo map + trim engine: header id (Phase 3 review) Addresses gemini-code-assist review on #4645: - Extract buildNamedComboLookup (pure) so a blank/whitespace/null combo name contributes only its id key (no '' key, no throw that disables all combos). - Trim the engine:<id> header value so 'engine: rtk' resolves. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Diego Rodrigues de Sa e Souza <diego.souza@cdwasolutions.com.br> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com> * fix: exclude exhausted connections from auto scoring (#4592) Integrated into release/v3.8.34 (rebuilt + opt-in gate fix) * fix(dashboard): memoize compatible provider groups (#4613) Integrated into release/v3.8.34 (rebuilt + test added) * fix(dashboard): isolate quota widget refresh clock (#4611) Integrated into release/v3.8.34 (rebuilt + jsdom test) * fix(dashboard): gate topology side effects behind widget visibility (#4606) Integrated into release/v3.8.34 (rebuilt + jsdom test) * fix(dashboard): keep play_arrow spinning on provider Test All buttons (#4563) Integrated into release/v3.8.34 (rebuilt onto tip; UI-cosmetic per owner) * fix(db): schedule retention cleanup + fix cleanup table/column names (extracted from #4428) (#4691) Integrated into release/v3.8.34 (cleanup core extracted from #4428, credit @oyi77) * fix(telemetry): back off live-WS event forwarding when the sidecar is unreachable (#4604) (#4687) Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com> * fix(api): serve GET /v1/models/{model} as JSON, not the HTML dashboard (#4674) (#4677) Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com> * feat(opencode): add go deepseek reasoning variants (#4647) Integrated into release/v3.8.34 * fix(executors): robust deepseek-web tool-call parsing and agentic context retention (#4644) Integrated into release/v3.8.34 * fix(cli): authenticate `omniroute logs` and honor active context (#4638) Integrated into release/v3.8.34 (authored by Rahul Sharma, AI co-author trailer stripped per project policy) * fix(proxy): apply pipelining:0 + connections cap to the direct dispatcher (#4580) (#4684) Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com> * fix(executors): Firecrawl web_fetch 500 with include_metadata=true (#4692) Integrated into release/v3.8.34 * fix(routing): include all noAuth models in auto-combos + add reka-flash + best-free template (#4621) Integrated into release/v3.8.34 (dead getFirstRegistryModelId dropped, rebuilt onto tip) * fix(dashboard): gate home topology live-WS networking (#4596) (#4618) Integrated into release/v3.8.34 (adapted onto #4606's extracted topology section: default-hidden flip + enabled gate on useLiveDashboard) * fix(cli): align `omniroute` env loading with the runtime data dir (#4597) (#4619) Integrated into release/v3.8.34 (data-dir.mjs refactor reconciled with #4607; loadEnvFile aligned to getDefaultDataDir) * chore(quality): reconcile file-size baseline for #4644 (deepseek-web.ts 1117->1125) (#4695) file-size reconcile for #4644 * Support quota scraping for OpenCode Go and Ollama Cloud (#4642) Integrated into release/v3.8.34 (Ollama Cloud + OpenCode Go dashboard quota scraping; rebuilt onto tip, gates green: typecheck/public-creds/file-size/lint/docs-sync + 31 tests) * feat(executors): land M365 Copilot pure framing + connection helpers (#4042) (#4696) Land M365 pure modules ahead of draft #4400 * deps: bump production + development groups; migrate js-yaml to v5 ESM (#4697) Incorporates Dependabot #4667 + #4668 + js-yaml v5 ESM migration into release/v3.8.34 * fix: noAuth provider validation + kimi executor routing (#4699) Integrated into release/v3.8.34 (noAuth in NOAUTH_PROVIDERS dynamic check + remove misrouted kimi web alias; 9 tests) * refactor(imageGeneration): extract 8 provider families to co-located files (#4609) Integrated into release/v3.8.34 (extraction completed: added missing imports/exports per module, main imports handlers locally; 145 image-gen tests pass, typecheck/cycles/file-size green) * chore(release): v3.8.34 — finalize changelog, rebaseline drift, fix release-green reds - Finalize CHANGELOG [3.8.34] (43 bullets, full contributor attribution) + seed i18n mirrors - Rebaseline inherited cycle drift surfaced by release-green pre-flight: eslint warnings 3900->3907, cognitive-complexity 797->801 (release-finalize touches no prod code; all drift is from this cycle's contributor merges) - fix(providers): keep reka-flash-3 as the Reka provider default. #4621 inserted reka-flash at the head of the model list, silently changing the default from reka-flash-3 (the free-tier model) to reka-flash; reorder so reka-flash-3 stays default, reka-flash retained. - test: align provider-models-config / provider-models-route / web-cookie-providers-new with #4621 (reka-flash now in the Reka catalog) and #4699 (the `kimi` API-key provider correctly falls through to DefaultExecutor instead of KimiWebExecutor) - chore(quality): allowlist the COMPRESSION_GUIDE doc name in check-fabricated-docs (false-positive env-var match; docs/compression/COMPRESSION_GUIDE.md exists) * fix(release-green): resolve release-PR full-CI reds for v3.8.34 Surfaced only on the release PR (these gates don't run on PR->release fast-gates): - fix(quota): complete HTML-comment sanitization in opencodeOllamaUsage SSR reset-time parsing — strip any <!--...--> generically instead of the two literal React hydration markers, so no partial "<!--" can survive (CodeQL js/incomplete-multi-character- sanitization, HIGH, introduced by #4642). Regression test added. - test(codex): correct the Codex-fingerprint body key order assertion to match the canonical bodyFieldOrder (prompt_cache_key precedes include); #4584 flipped the two and integration tests don't run on fast-gates so it never executed until the release PR. - chore(quality): rebaseline inherited cycle drift surfaced by full CI — zizmorFindings 152->155 (+3 unpinned-uses in nightly-release-green.yml from #4622, same @vN convention as ci.yml) and openapiCoverage.pct 38.4->37.8 (-0.6, contributor routes added faster than openapi docs). Release-finalize touches no prod routes. * fix(release-green): complete CodeQL sanitization + rebaseline complexity drift - fix(quota): handle unterminated HTML comments in opencodeOllamaUsage SSR reset-time parsing — the `(?:-->|$)` arm consumes a trailing "<!--" with no closing "-->", so no partial "<!--" can survive (CodeQL js/incomplete-multi-character-sanitization persisted with the plain <!--...--> form because an unclosed comment could still leave "<!--"). - chore(quality): rebaseline cyclomatic complexity 1915->1916 (+1) — inherited v3.8.34 cycle drift (contributor feature branches); check:complexity does not run on PR->release fast-gates so it surfaced only on the release PR. Release-finalize adds 0 complexity (measured 1916 with/without the regex tweak). dead-code/cognitive/type-coverage/ compression-budget/codeql ratchets all pass. --------- Co-authored-by: Diego Rodrigues de Sa e Souza <diego.souza@cdwasolutions.com.br> Co-authored-by: Randi <55005611+rdself@users.noreply.github.com> Co-authored-by: Hernan Javier Ardila Sanchez <hjasgr@gmail.com> Co-authored-by: KooshaPari <42529354+KooshaPari@users.noreply.github.com> Co-authored-by: Abhishek Divekar <adivekar@utexas.edu> Co-authored-by: Rahul sharma <sharmaR0810@gmail.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com> Co-authored-by: Ronald Estacion <DevEstacion@users.noreply.github.com> Co-authored-by: Igor <60442260+BugsBag@users.noreply.github.com> Co-authored-by: Oonishi <275808243+ponkcore@users.noreply.github.com> Co-authored-by: Paijo <14921983+oyi77@users.noreply.github.com> Co-authored-by: Jan Leon <Jan.gaschler@gmail.com> |
||
|
Diego Rodrigues de Sa e Souza
|
41eb0091a2 |
fix(sse): parse <tool_call name=...> wrapper from web-cookie providers (#3260) (#3275)
ds-web/deepseek-v4-pro emits tool calls wrapped as
<tool_call name="skill">{"name":"customize-opencode"}</tool_call> instead of the
canonical <tool>{json}</tool>. webTools.ts only matched <tool>...</tool>, so the block
was silently dropped (and when arguments were present, the surrounding tag leaked into
content). Add TOOL_CALL_TAG_RE to capture the JSON body — the real tool name comes from
the body, never the tag's name= attribute — and extend the early-exit + range stripping.
Regression test: tests/unit/web-tools-translation-3260.test.ts (RED before, GREEN after).
Existing web-tools suites stay green (26/26).
|
||
|
Paijo
|
e364764dc7 |
feat(web-cookie): add tool-call translation to 8 executors via shared webTools helpers (#3259)
Add tool-call translation to 8 web-cookie executors via shared webTools helpers. Integrated into release/v3.8.12. Thanks @oyi77. |
||
|
Diego Rodrigues de Sa e Souza
|
68d5a0ab27 |
Release v3.8.10 (#3140)
* chore(release): open v3.8.10 development cycle Bump 3.8.9 → 3.8.10 across package.json, lockfile, electron, open-sse, and docs/reference/openapi.yaml; add the [3.8.10] CHANGELOG section (root + 41 i18n mirrors) as the integration target for the cycle. Entries land here as work merges into release/v3.8.10; finalized by the release flow. * fix(providers): resolve web provider alias collisions Assign unique aliases to HuggingChat, Kimi Web, and Qwen Web so they no longer shadow primary providers or trigger startup warnings. Add a unit test to enforce provider alias uniqueness and prevent future collisions. Also expand local ignore and VS Code exclude rules for agent, build, and worktree artifacts. * fix(responses): normalize image_url parts across input paths (#3150) Normalize image_url parts across all Responses input paths. Integrated into release/v3.8.10. * fix(api-manager): preserve API key expiration local time (#3146) Preserve API key expiration local time + clear button. Integrated into release/v3.8.10. * Strip previous_response_id for stateless Responses upstreams (#3143) Strip previous_response_id for stateless Responses upstreams (auto/strip/preserve). Integrated into release/v3.8.10. * fix(opencode-plugin): map thinking cap to interleaved in model+combo (#3138) Map caps.thinking to ModelV2.capabilities.interleaved for opencode-plugin. Integrated into release/v3.8.10. * fix(providers): use synced models as fallback for all providers (#3148) Use synced models as authoritative local catalog for all providers (+regression test). Integrated into release/v3.8.10. * fix(qoder): bifurcate validation by token type — PAT→Cosy, regular API key→dashscope (#3149) Bifurcate Qoder validation by token type (PAT→Cosy, regular→dashscope) +regression test. Integrated into release/v3.8.10. * fix(antigravity): dynamic model resolution via MITM alias table (#3144) Dynamic antigravity MITM model resolution in the executor (+bug fix +regression test; DB import dropped from client-reachable config). Integrated into release/v3.8.10. * Feature/batch allow big (#3128) Podman deployment options + larger upload body-size limits (+CONTAINER_HOST docs). Integrated into release/v3.8.10. * fix(fireworks): preserve fully-qualified router/model IDs (#3133) (#3160) Fireworks router IDs (accounts/fireworks/routers/...) were double-prefixed with accounts/fireworks/models/ → upstream 404. Add optional acceptedModelIdPrefixes to the registry entry and skip the prepend when the model already starts with an accepted prefix. Co-authored-by: KooshaPari <KooshaPari@users.noreply.github.com> * fix(llama-cpp): route to configured local baseUrl instead of OpenAI (#3136) (#3161) llama-cpp was missing from the local-provider group in buildUrl(), so it fell through to the OpenAI baseUrl and returned an OpenAI 401. Add the case to resolve the connection's providerSpecificData.baseUrl. Co-authored-by: tjengbudi <tjengbudi@users.noreply.github.com> * fix(t3-chat-web): parse cookies + convexSessionId from stored credential (#3007) (#3162) The executor read credentials.cookies/convexSessionId, but the pipeline only stores the pasted string under apiKey → t3.chat always 400'd. Parse both values from apiKey (fallback accessToken), mirroring validation.ts. Co-authored-by: minhtran162 <minhtran162@users.noreply.github.com> * fix(minimax): stop capping MiniMax-M3 / M2.7 max_tokens at 8192 (#3141) (#3163) MiniMax-M3 had no MODEL_SPECS entry and capitalized MiniMax-M2.7 missed its lowercase spec (case-sensitive lookup) → both fell to the 8192 default cap. Add the M3 spec (512K output), alias the capitalized ids, and make getModelSpec lookups case-insensitive. Co-authored-by: totaltube <totaltube@users.noreply.github.com> * fix(github-copilot): discover model catalog live from api.githubcopilot.com (#3120, #3121) (#3164) The github (Copilot) provider had a static hardcoded catalog with no discovery source, so Import Models never refreshed (#3120) and advertised non-entitled models that 400 on use (#3121). Add a live /models fetch with fallback to the static list. Co-authored-by: gabrielmoreira <gabrielmoreira@users.noreply.github.com> * fix(combo): invalidate nested-combo cache on edits + log DATA_DIR (#3147) (#3165) Editing a combo did not invalidate the 10s nested-combo expansion caches (chat.ts getCombosCachedForChat + chatCore.ts getCombosCached; the exported clearCombosCache was dead code), so a removed nested target/model could be served as a phantom for up to 10s. Wire a shared monotonic combos-cache version in readCache (bumped by invalidateDbCache("combos") on every combo write); both cache layers treat a version mismatch as a miss. Also log the resolved DATA_DIR/SQLITE_FILE absolute path at DB init so the reporter's 'persists across restart + volume wipe' symptom (a multi-replica Docker volume/DATA_DIR mismatch, not a routing bug) is diagnosable from logs. Includes consolidated CHANGELOG entries for #3133/#3136/#3007/#3141/#3120/#3121. Co-authored-by: ViFigueiredo <ViFigueiredo@users.noreply.github.com> * fix(web-tools): parse bare JSON tool calls (#3157) Parse bare JSON tool calls for deepseek-web (#2820) + fuzzy tool-name matching. Integrated into release/v3.8.10. * fix(misc): minor fixes across reasoning cache, account fallback, binary manager (#3177) Misc: ProviderProfile export, DeepSeek reasoning regex, binary guard. Integrated into release/v3.8.10. * fix(kiro): minor OAuth social exchange tweaks (#3176) Kiro social OAuth: optional targetProvider passthrough. Integrated into release/v3.8.10. * deps: bump hono from 4.12.18 to 4.12.23 (#3179) Bump hono to 4.12.23. Integrated into release/v3.8.10. * fix(providerRegistry): update kilocode format and executor (#3166) kilocode: openai format + default executor (matches kilo-gateway) + registry test. Integrated into release/v3.8.10. * feat(metrics): cross-request TTFT and gap latency after tool calls (#3173) Cross-request TTFT + gap-after-tool latency metrics (+test). Integrated into release/v3.8.10. * feat(dashboard): provider stats API endpoint and dashboard page (#3175) Provider stats dashboard + API (SQL moved to db module per Hard Rule #5, +test). Integrated into release/v3.8.10. * fix(usage): sequential+spaced OAuth quota sync, reactive force-refresh, actionable 401 (#3156) Sequential+spaced OAuth quota sync, reactive force-refresh on 401, actionable 401 in UI. Integrated into release/v3.8.10. * fix(healthcheck): per-provider proactive-refresh skip list (rescue short-TTL OAuth) (#3159) Per-provider proactive-refresh skip list (OMNIROUTE_HEALTHCHECK_SKIP_PROVIDERS) to rescue short-TTL OAuth. Integrated into release/v3.8.10. * feat(quota): show OAuth token expiry on provider cards (small, blue, informative) (#3178) Show OAuth token expiry on provider cards (small, blue, informative). Integrated into release/v3.8.10. * fix(providers): empty refresh must not resurface just-cleared synced models (#3181) Empty refresh must not resurface just-cleared synced models (fixes the release-blocking provider-models-route test). Integrated into release/v3.8.10. * chore(release): v3.8.10 — 2026-06-04 (finalize CHANGELOG) --------- Co-authored-by: Wilson <pedbookmed@gmail.com> Co-authored-by: Xiangzhe <32761048+xz-dev@users.noreply.github.com> Co-authored-by: Jan Leon <Jan.gaschler@gmail.com> Co-authored-by: M.M <mr.maatoug@gmail.com> Co-authored-by: Hernan Javier Ardila Sanchez <hjasgr@gmail.com> Co-authored-by: Markus Hartung <mail@hartmark.se> Co-authored-by: KooshaPari <KooshaPari@users.noreply.github.com> Co-authored-by: tjengbudi <tjengbudi@users.noreply.github.com> Co-authored-by: minhtran162 <minhtran162@users.noreply.github.com> Co-authored-by: totaltube <totaltube@users.noreply.github.com> Co-authored-by: gabrielmoreira <gabrielmoreira@users.noreply.github.com> Co-authored-by: ViFigueiredo <ViFigueiredo@users.noreply.github.com> Co-authored-by: PizzaV <103120356+pizzav-xyz@users.noreply.github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Nicolas Lorin <androw95220@gmail.com> |
||
|
Diego Rodrigues de Sa e Souza
|
50896699d3 |
fix+feat(sse): per-model 403 lockout (#3027) + deepseek-web memory (#2942) & tool-calls (#2820) (#3101)
* fix(sse): per-model 403 on passthrough providers locks the model, not the connection (#3027)
A per-model subscription 403 from a passthrough / per-model-quota provider
(e.g. ollama-cloud "this model requires a subscription, upgrade for access" on
deepseek-v4-pro) cooled down the ENTIRE connection instead of locking out only
the paid model, knocking out the free models on the same key and escalating an
exponential connection-wide backoff on repeats.
markAccountUnavailable's per-model lockout gate only covered 404/429/>=500, and
the only 403 -> model-lockout special case was hard-coded to grok-web. Generalize
it: a model-scoped 403 on an isPerModelQuotaProvider becomes a model lockout
(connection stays active). Terminal whole-key 403s (permanent/banned, account
deactivated, credits exhausted, project-route) keep their connection-level path.
Test-first: reproduction (paid model locked, connection active, free model still
eligible), regression guard (deactivated key still terminal -> banned, not
downgraded), and backoff guard (repeated 403s do not escalate connection backoff).
* feat(sse): persistent session + rolling-window memory for deepseek-web (#2942)
The DeepSeek web API takes only a single `prompt` string (no messages array) and
the executor created a fresh chat session per request, deleting it afterward — so
agentic multi-turn clients got per-turn amnesia.
Add two opt-in, per-connection settings (providerSpecificData), both defaulting to
the legacy behavior so plain-chat users are unaffected:
- historyWindow (number, default 0): when > 0, messagesToPrompt stitches the last N
non-system turns into a role-tagged transcript ("User:"/"Assistant:") so context
carries across turns within the single prompt string.
- persistSession (bool, default false): reuse one upstream chat session per userToken
(cached in the existing sessionCache) instead of creating/deleting one per request.
A reused session that fails is treated as stale (deleted in the DeepSeek UI): the
cache entry is dropped, a fresh session is created, and the completion is retried
once. Error paths invalidate the cached session so the next turn self-heals.
Test-first: pure prompt-builder window semantics (legacy/window/cap/empty) and
execute()-level session behavior via mocked fetch (fresh-per-request default, reuse,
stale-session fresh retry, history threaded into the prompt). Full existing
deepseek-web suite (35 tests) still green.
Note: chat.deepseek.com is an unofficial reverse-engineered surface; the live
round-trip can't be exercised in CI (no userToken/session). Treated as best-effort
per the issue; the prompt/session logic is unit-tested in isolation.
* feat(sse): tool-call translation for deepseek-web (#2820)
deepseek-web previously hard-failed any request carrying tools[] with a 400 (#2848),
so agentic clients could not use it for function calling at all. Add a bidirectional
translation layer (new open-sse/translator/webTools.ts, reusable by the other
web-cookie executors):
- Request: serializeToolsToPrompt() turns the OpenAI tools[] into a <tool>{...}</tool>
prompt contract, injected as a leading system message.
- Response: parseToolCallsFromText() extracts the model's <tool> blocks into OpenAI
tool_calls (arguments as a JSON string), strips them from content, and the executor
emits finish_reason "tool_calls" — for both non-stream and stream clients (the reply
is buffered since a tool block must be parsed whole).
A plain reply (no <tool> block) still streams normally with finish_reason "stop".
The superseded #2848 400-contract test is updated to assert the new translation.
Test-first: pure serializer/parser units + execute() round-trip via mocked fetch
(no-400, prompt serialization, non-stream tool_calls, stream tool_calls, plain reply).
Note: chat.deepseek.com is an unofficial reverse-engineered surface and the exact
<tool> emission depends on the model following the injected contract; the live
round-trip can't be exercised in CI. Best-effort per the issue; translation logic is
unit-tested in isolation.
* fix(sse): drop duplicate per-model 403 block — already in release via #3096
The release branch already scopes per-model 403 to model lockout (PR #3096,
commit
|