Skip to content

feat: auto-promote non-repair candidate skills#2009

Draft
RerankerGuo wants to merge 1 commit into
MemTensor:mainfrom
RerankerGuo:feat/auto-promote-candidate-skills
Draft

feat: auto-promote non-repair candidate skills#2009
RerankerGuo wants to merge 1 commit into
MemTensor:mainfrom
RerankerGuo:feat/auto-promote-candidate-skills

Conversation

@RerankerGuo

Copy link
Copy Markdown

Fixes #1941.

Summary

  • Auto-promote untried non-repair candidate skills once their eta reaches the retrieval threshold.
  • Keep repair-origin candidates on the trial path so repairs still require evidence before activation.
  • Run a skill lifecycle tick during pipeline flush after pending skill writes are drained.

Validation

  • npm test -- --run tests/unit/skill/lifecycle.test.ts
  • npm run lint
  • Attempted: npm test -- --run tests/unit/skill/subscriber.test.ts -t "lifecycleTick"; blocked before assertions because better-sqlite3 native bindings are unavailable in this local Node 18 / compiler environment.
  • Attempted: make format; blocked because poetry is not installed in this environment.

@Memtensor-AI Memtensor-AI changed the base branch from main to dev-v2.0.22 July 1, 2026 13:14
@CarltonXiang CarltonXiang deleted the branch MemTensor:main July 3, 2026 07:25
@syzsunshine219 syzsunshine219 reopened this Jul 3, 2026
@syzsunshine219 syzsunshine219 added the needs-audit Requires manual audit before merge label Jul 3, 2026
@syzsunshine219 syzsunshine219 changed the base branch from dev-v2.0.22 to main July 3, 2026 08:20
@RerankerGuo

Copy link
Copy Markdown
Author

Closing this draft because current main already contains the candidate skill lifecycle promotion path, including the lifecycle tick after skill flush.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

needs-audit Requires manual audit before merge

Projects

None yet

Development

Successfully merging this pull request may close these issues.

feat: auto-promote non-repair candidate skills without trial

3 participants