Darwin Skill

Role In The Wiki

darwin-skill is a local D-drive repository for optimizing agent skills through iterative evaluation, improvement, testing, human confirmation, and git-controlled retention or rollback.

Current Claims

  • EXTRACTED: The repository tracks https://github.com/appleweiping/darwin-skill.git on branch master.
  • EXTRACTED: The original upstream repository is https://github.com/alchaincyf/darwin-skill.git.
  • EXTRACTED: At the 2026-05-16 trace, local HEAD matched upstream HEAD at 2056abfccd924d68ae6baa9193cafff0f666260b.
  • EXTRACTED: Public GitHub metadata described alchaincyf/darwin-skill as an autoresearch-inspired autonomous skill optimization system.
  • EXTRACTED: The English README describes the project as “Optimize your Agent Skills the way you train models.”
  • EXTRACTED: The core loop is evaluate, improve, test, human confirm, keep or revert, repeat.
  • EXTRACTED: The project is inspired by Karpathy’s autoresearch and maps autonomous experiment loops onto SKILL.md optimization.
  • EXTRACTED: The root SKILL.md describes an 8-dimension rubric covering structure and real-world effectiveness.
  • INFERRED: This repository is relevant for auditing, scoring, and improving existing skills rather than merely collecting them.

Local Routing

  • Local root: D:/Skill/darwin-skill.
  • Git remote: https://github.com/appleweiping/darwin-skill.git.
  • Original upstream: https://github.com/alchaincyf/darwin-skill.
  • Branch at inspection: master.
  • Use for: skill review, skill scoring, skill optimization, result-card generation, and skill-quality experiments.