alchaincyf

alchaincyf / darwin-skill

Public

达尔文.skill —— 一个让你的Skill无限进化的系统：评估→改进→测试→保留或回滚 | Autoresearch-inspired autonomous skill optimization for Claude Code. Evaluate, improve, test, keep or revert.

994
115
100% credibility
Found Apr 13, 2026 at 66 stars 15x -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
HTML
AI Summary

A Claude Code skill that iteratively optimizes other skills by evaluating structure and performance, applying targeted improvements, and retaining only those that increase the score with human confirmation.

How It Works

1
🔍 Discover Darwin Skill

You hear about a helpful tool that automatically makes your Claude Code assistants smarter and better over time.

2
➕ Add It to Your Toolbox

You easily add darwin-skill to your collection of helpers in Claude Code.

3
Choose What to Improve
🌐
Improve All

Let it evolve your entire set of helpers automatically.

📋
Pick One

Target a specific helper that needs a boost.

4
🔄 Watch the Evolution Loop

It checks performance, tries smart tweaks, scores the results, and only keeps winners that score higher.

5
✅ Review and Approve

You take a quick look at the changes and differences, then give the go-ahead if they look good.

🚀 Smarter Assistants Ready

Your helpers are now sharper, more effective, and keep getting better with each round.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 66 to 994 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is darwin-skill?

darwin-skill is an autoresearch-inspired autonomous optimizer for Claude Code skills. It runs darwin skill assessment on your SKILL.md files—scoring structure and live output across 8 dimensions—then targets weaknesses to improve, test, and keep gains or revert via a ratchet mechanism. Install as an HTML-based Claude skill with `npx skills add alchaincyf/darwin-skill`, then say "optimize all skills" to evolve your darwin skills collection.

Why is it gaining traction?

It stands out by blending full autonomy with human-in-the-loop checks, ensuring darwin skills development only advances your baseline score (never regresses) through dual evaluation of format and effectiveness. Developers hook on the darwin optimization loop: find low scores, propose fixes, validate on test prompts, and commit only winners—mirroring model training but for Claude Code. No alternatives offer this targeted, revert-safe darwin skill assessment for scaling 60+ skills.

Who should use this?

Claude Code heavy users managing darwin skilled lists of custom skills for code gen, debugging, or workflows. Ideal for AI prompt engineers building skill ecosystems who hate manual tweaks, or teams iterating darwin skills development schemes for consistent output quality. Skip if you're under 10 skills or not deep in Claude.

Verdict

Worth adding for Claude Code enthusiasts—clever ratchet keeps darwin skills sharp—but at 43 stars and 1.0% credibility, it's early-stage with solid docs yet unproven at scale. Test on a few skills first; pair with nuwa-skill for full evolution.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.