Evaluate and improve Claude-style SKILL.md entries with Git-backed optimization.
npx skills add https://github.com/wukangcheng2944/claude-skills --skill darwin-skill
Please help me install this Agent Skill. Skill: darwin-skill Source: https://github.com/wukangcheng2944/claude-skills/tree/main/darwin-skill Command: npx skills add https://github.com/wukangcheng2944/claude-skills --skill darwin-skill
Darwin.skill provides an automated framework for evaluating and evolving Claude-style SKILL.md entries. It orchestrates a rigorous, repeatable optimization loop that examines both the static structure of SKILL.md and the real-world effectiveness of its guidance, ensuring skills actually improve over time rather than just looking better on paper.
Install darwin.skill and start the optimization workflow by telling your agent to optimize all skills.
High-intent search queries and answers about installing and using this skill.
Automating Claude-style SKILL.md evaluation requires running baseline scoring, generating improvements, live testing with designed test prompts, and using human-in-the-loop confirmation within a Git-backed workflow to ensure only measurable improvements are kept.
Git-backed hill-climbing optimization for agent skills is a workflow that applies improvements through commits and automatically reverts changes if live testing and rubric scoring fail to show measurable effectiveness gains over the baseline.
An eight-dimension rubric scores agent skill effectiveness by evaluating static structure for 60 points and real-world effectiveness for 40 points, using independent scoring agents to ensure skills actually improve rather than just looking better on paper.
Human-in-the-loop confirmation allows you to review SKILL.md optimization cycles by pausing after each iteration to generate diff summaries and visual result cards, ensuring you approve only verified improvements before they are committed.
Playwright is required to run live testing on agent skills, serving as the core dependency for executing designed test prompts against the optimized SKILL.md entries to validate their real-world effectiveness before committing changes.
Optimization loops fail to keep improvements when changes do not measurably outperform baseline scores in the eight-dimension rubric or live test prompts, triggering an automatic Git revert to discard the unverified modifications.
rodrigo-altissimoRank solution candidates using weighted scoring and TCO-based convenience premiums.
MichaelSolatiCompare baseline and current GEO audit files to generate markdown progress reports.
Spectra29115Execute automated end-to-end UI tests with Playwright and capture screenshots.
pingqLINAnalyze SKILL.md files with an 8-dimension rubric and optimize via git-backed hill-climbing.
