Commit Graph
10 Commits
Author SHA1 Message Date
xuhdandClaude Opus 4.7 c493639119 feat(skill-review): strict-mode prompt + raise min_tool_calls 3 to 5
The review agent was creating too many low-value skills. Three changes
tighten the funnel:

- review_worker prompt: rewritten in "保守模式". Default decision is
  do-nothing. To create, all five AND conditions must hold (>=5 tool
  calls with step dependencies, an explicit fail->fix->success episode,
  cross-scenario reusability, no >=70% overlap with skill_list, can
  produce >=300 chars of substance). Adds 7 explicit NO categories
  (single-tool loops, one-off queries, LLM common knowledge, etc.)
  and a 4-question self-check before any create. Hard cap of 1 new
  skill per review, with patch preferred over create.
- min_tool_calls: 3 -> 5. Matches the prompt's ">=5 steps" hard
  condition so trivial turns never reach the LLM reviewer.
- Updated config.yaml, the default-value assertion in
  test_skill_review_worker, and the middleware docstring's "(default
  3)" annotation to keep all three references in sync.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-06-09 16:00:29 +08:00
xuhd 6f9afcdba2 修复技能审核agent提示词 2026-06-05 13:32:19 +08:00
xuhd 705de81c78 fix(skill-review): use uuid per review run to avoid checkpoint context bleed 2026-06-05 13:32:19 +08:00
xuhd 335bc3cf48 fix(skill-review): propagate user_id through review chain for correct DB ownership 2026-06-05 13:32:19 +08:00
xuhd 857798ff2f fix(skill-review): pass RunnableConfig with thread_id to satisfy ThreadDataMiddleware 2026-06-05 13:32:19 +08:00
xuhd 3955bd72b5 refactor(skill-review): use create_agent with skill_evolution disabled to allow autonomous skill creation 2026-06-05 13:32:19 +08:00
xuhd b5d80025e9 fix(skill-review): lower CREATE threshold and log agent decision 2026-06-05 13:32:19 +08:00
xuhd ad5bd22c5e refactor(skill-review): use make_lead_agent instead of create_react_agent
Reuses the full middleware chain (ToolErrorHandlingMiddleware,
LLMErrorHandlingMiddleware, etc.) so tool errors like 'already exists'
and invalid SKILL.md are converted to ToolMessages by the middleware,
not silently swallowed or crashing the review task.
2026-06-05 13:32:19 +08:00
xuhd b2ed6f1acf fix(skill-review): add strict SKILL.md format spec to review prompt 2026-06-05 13:32:19 +08:00
xuhdandClaude Sonnet 4.6 ef3ef433c1 feat(skill-evolution): add SkillReviewWorker for passive post-turn skill creation
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-06-05 13:32:19 +08:00