Commit Graph

9 Commits

Author SHA1 Message Date
matevip
82538ca3aa feat(llm): add Claude Opus 4.8 + 4.8 Fast model entries 2026-05-29 07:06:19 +08:00
matevip
9e93c52d9a fix(goal): real evaluator, retry refactor, hardened node + extra edges 2026-05-21 22:27:00 +08:00
matevip
c34e8290ac feat(goal,ui): inline set-goal prompt, terminal system-line, sidebar dot 2026-05-21 22:26:40 +08:00
matevip
9c5ad29d42 chore(llm): drop unused imports, add XIAOMI_MIMO cross-turn cache tests 2026-05-21 17:24:53 +08:00
倪程伟
7861f603eb
fix(llm): MiMo thinking 模式 reasoning_content 多轮对话兼容修复 (#189)
MiMo V2 系列在 thinking 模式下,assistant 消息携带 tool_calls 时必须同时包含 reasoning_content,否则提供方返回 400。

- ModelFamily 新增 MIMO_THINKING 族,detect() 添加 mimo* 匹配
- FallbackPolicy 新增 XIAOMI_MIMO(patchCrossTurn=true, patchNonToolCall=true)
- 新增 ReasoningContentCache,按 tool_call_ids 回放真实推理内容
- 缓存作用范围:所有 patchCrossTurn=true 的 thinking provider(MiMo + DeepSeek)
- NodeStreamingChatHelper 流式响应完成后写入缓存

Closes #188
2026-05-21 17:15:35 +08:00
matevip
75107ac815 feat(llm): native Gemini chat builder, Nano Banana image gen, xAI/Grok provider 2026-05-18 10:00:55 +08:00
matevip
a88edbdd07 refactor(llm): decouple model construction from the agent graph layer (#147) 2026-05-18 07:47:49 +08:00
matevip
55f4ba1195 feat(llm): per-Model HTTP read-timeout override 2026-05-02 15:40:59 +08:00
matevip
3b11a3def6 fix(failover): AUTH_ERROR triggers fallback chain + UI splits provider 401 from session expiry
Two related issues from the Kimi-401 user report:

1. Backend (NodeStreamingChatHelper): a primary AUTH_ERROR (e.g. Kimi 401
   with an invalid API key) returned immediately without trying the
   fallback chain — a fallback provider with a different, valid key
   never got a chance. Even with DashScope correctly configured as the
   fallback, the user chat dead-ended on a 401.

   The original assumption ("auth never self-heals so do not retry")
   holds for the primary same-model retry loop but is wrong for the
   fallback chain — different providers have different keys. Apply the
   same break-into-fallback policy that BILLING and MODEL_NOT_FOUND
   already use. recordPrimary(false) is preserved so the cooldown
   counter still accumulates.

2. Frontend (chatError.ts + i18n): the error-text matching for
   /认证|auth|unauthorized|401/i was so broad it matched the substring
   "auth" inside URLs like https://api.kimi.com/.../auth, classifying
   any model 401 as user "session expired" and rendering the misleading
   "页面将自动跳转到登录页" copy. (The redirect itself only fires from
   /api/v1/auth/* axios paths and SSE-connection 401s, not from this
   payload-text path — but the copy alone is the worst kind of false
   alarm.)

   Add a new ChatErrorCategory provider_auth_error and split the
   pattern matching: narrow auth_expired (HTTP 401 / 登录已过期 /
   session expired / 凭证失效) is matched FIRST, then the broad
   401-ish pattern routes to provider_auth_error. BACKEND_ERROR_TYPE_MAP
   for AUTH_ERROR is also remapped, since structured backend payloads
   currently always come from LLM providers — never from our own
   /api/v1/auth path.

Tests
- NodeStreamingChatHelperFailoverTest (5 cases): primary 401 →
  fallback succeeds; chain skips auth-failing fallback to next healthy
  one; whole-chain failure surfaces last AUTH_ERROR (no silent drop);
  BILLING regression unchanged; primary-success path does not touch
  chain
- Browser preview verified: new i18n keys resolve in en-US, classifier
  correctly routes "[错误] 401 from kimi.com" → provider_auth_error
  while "[错误] HTTP 401 from /api/v1/auth/ping" stays auth_expired
- 186 tests pass (was 181 + 5 new); vue-tsc clean

Do-not-touch list: handleAuthFailure() in useStream/api/index.ts (real
session-expiry path) is unmodified — only the misclassification
upstream is fixed. auth_expired i18n copy is unchanged.
2026-04-19 17:45:15 +08:00