Darwin Skill 2.0 Evolution System
Paste the following prompt into your AI chat to install this skill:
Please install @user_15292d5a/yjkj-darwin-skill-qszf using https://skillhub.cn/install/skillhub.md.
About this skill
Problem
When a SKILL.md already works but has loose structure, ambiguous steps, or missing error boundaries, manual edits can become guesswork. darwin-skill-qszf targets iteration on an existing Skill by turning optimization into a reviewable loop: score it, change it, test it, and keep or roll back the result.
How It Works
The skill treats one SKILL.md as the single editable asset and optimizes one Skill at a time. The main workflow is:
- Scoring: A subagent independently scores nine dimensions, including structural completeness, clarity, content completeness, operability, accuracy, consistency, execution efficiency, robustness, and meta-skill compliance.
- Improvement: It adjusts low-scoring areas such as instructions, steps, paths, boundary conditions, and tool guidance, avoiding unrelated rewrites.
- Testing and validation: test-prompts verify behavior, while an independent judge agent performs blind evaluation to reduce self-editing bias.
- Keep or roll back: A ratchet mechanism preserves only measurable improvements; if validation gets worse, the skill reverts to the previous version.
- Human in the loop: Each optimization round pauses for user confirmation before finalization, and can produce a result card.
Boundaries
It fits existing SKILL.md files that need structured optimization, effect validation, and version control. For creating a Skill from scratch, the source suggests pairing it with nuwa-skill. Use it with runnable test prompts and keep a check on rollback points.
Use Cases
- An existing `SKILL.md` is loose or ambiguous, so you need nine-dimension scoring to find weak areas and rewrite instructions and boundaries.
- After changing an Agent skill prompt, you need `test-prompts` and blind judging to verify improvement before keeping or rolling back.
- A team maintains several skills and needs a pause-and-confirm step after each optimization round to preserve reviewable changes.
- When creating or maintaining AI Agent skills, you need to separate creation from optimization and run an iteration loop.
Best For
- Engineers maintaining Agent skill files who need to turn scattered `SKILL.md` files into clear, executable versions.
- Applied engineers optimizing prompts who need low-score diagnosis, test validation, and automatic rollback.
- Teams using Nuwa to create skills who need follow-up iteration and version-quality control.
- Architects reviewing Agent workflows who need independent scoring and human-in-the-loop confirmation.
Related Skills
A free matrix publishing assistant for Skill developers, guiding distribution to Tencent SkillHub, XiaLiao, XiaYou SkillHub, GitHub, and ClawHub with create, update, and delete safeguards.
A guide for writing Seedance 2.0 video prompts, covering @ references, camera moves, time segments, audio direction, and common pitfalls.
Collects daily conversations, task logs, and errors, then runs a nine-module review to surface blind spots, generate actions, and update memory.
Local long-term memory for AI agents that appends key facts and daily logs, supports cross-session context and search recall, and auto-archives old diaries to gzip files.