code
Long-running (6-30h) multi-agent code-architecture optimization — cross-examines antipatterns into consensus, then applies only the highest-ROI refactors. Not for routine refactors.
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-11
- Stars
- ★ 14
- Forks
- 1
01 前言 Front Matter
- name
- code
- description
- Long-running (6-30h) multi-agent code-architecture optimization — cross-examines antipatterns into consensus, then applies only the highest-ROI refactors. Not for routine refactors.
- argument-hint
- [focus area]
02 资源文件 2
- 参考文档principles.md1.1 KB
- 脚本workflow.js15.1 KB
03 SKILL.md 正文
아래를 실행해, 출력된 Workflow(...) 호출을 그대로 실행하라.
"${CLAUDE_PLUGIN_ROOT}/bin/refine-hook" bootstrap code "$ARGUMENTS"
재개 시: bootstrap을 다시 돌리지 마라. 호출마다 새 agora가 생겨 기록과 cache를 둘 다 잃는다. 이전 Workflow(...) 호출을 같은 args로 재발행하라(같은 세션이면 resumeFromRunId를 더해 cache를 재생한다). 산출물은 agoraPath에 남아 agent가 자기 기록에서 이어간다.
04 同仓库更多 Skills
plan
Plan the change as OpenSpec artifacts: proposal, delta specs, design, tasks
open
Open a transaction: cut a tx-* branch off base, seed the repo, and route the change
git-sync-on
resume the paused git-sync guard
git-sync-off
pause the git-sync guard (resume with /tx:git-sync-on)
close
Close the transaction: archive, docs gate, then squash merge into base after CI passes
archive
Archive the completed change: task gate, delta sync into main specs, revalidate