skills-repo-health-check
Skill14 starsChanged 5 months agoMITView on GitHub
Audit and improve Agent Skills repositories. Use when the user asks to check a skills repo, assess skill quality, find duplicate or overlapping skills, review SKILL.md/frontmatter/description quality, verify README or plugin index sync, compare against high-quality skill repository practices, or produce a prioritized plan to evolve a skills collection.
In this repository39 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- logo-generatorSkill
- source-command-kj-helpSkill
- ai-coding-weekly-reportSkill
- ai-conversation-to-wikiSkill
- ai-daily-from-xSkill
- ai-daily-websearchSkill
- ai-learning-loopSkill
- ai-programming-topic-plannerSkill
- ai-reading-coachSkill
- coding-session-to-tutorialSkill
- content-creatorSkill
- content-growth-reviewSkill
- content-illustratorSkill
- course-campaignSkill
- course-generatorSkill
- course-scraperSkill
- english-reading-studioSkill
- extract-conversation-insightsSkill
- git-pushSkill
- humanizer-zhSkill
- industry-best-practicesSkill
- kangjian-douyin-mediaSkill
- kangjian-skillSkill
- knowledge-mapSkill
- markdown-formatSkill
- paid-content-reviewSkill
- phenomenon-insightSkill
- research-backed-content-loopSkill
- story-pptSkill
- transcript-critical-analysisSkill
- tutorial-guideSkill
- web-slideSkill
- wechat-companionSkill
- wechat-publish-kitSkill
- weekly-digestSkill
- wiki-doc-sinkSkill
- wiki-intelligenceSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Use this skill for one-off audits, pre-release checks after adding or renaming skills, migration/de-duplication reviews, and periodic …
- Distinctive: line 17
Run the quick audit script when possible:
- Very specific: line 23
Use --json when another tool will consume the result.
- Very specific: line 25
Read references/best-practices.md before producing the final assessment.
- Distinctive: line 27
Check repo-level discovery surfaces: README, contributor guide, plugin manifests, marketplace files, help commands, sync scripts, and …
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Skills Repo Health Check
Overview
Audit a skills repository as a capability system, not a folder of prompts. Combine automated checks with judgment about trigger quality, workflow design, verification, discoverability, safety, and repo governance.
Use this skill for one-off audits, pre-release checks after adding or renaming skills, migration/de-duplication reviews, and periodic quality reviews.
Workflow
- Identify the repository root and skill directories. Prefer
skills/<skill-id>/SKILL.md; if another layout is present, verify the real implementation path before judging. Run the quick audit script when possible:
python3 skills/skills-repo-health-check/scripts/audit_skills_repo.py .Use
--jsonwhen another tool will consume the result.- Read
references/best-practices.mdbefore producing the final assessment. Use it as the scoring rubric and improvement checklist. - Inspect a representative set of weak or risky skills directly, especially any skill flagged for vague triggers, missing verification, long
SKILL.md, missingagents/openai.yaml, or possible overlap. - Check repo-level discovery surfaces: README, contributor guide, plugin manifests, marketplace files, help commands, sync scripts, and platform-specific docs.
- Produce a decision-ready report with scores, P0/P1/P2 issues, concrete file paths, and a staged improvement plan.
Audit Dimensions
Score each dimension from 0-10:
- Structure consistency: canonical folders, required
SKILL.md, optional resources only when useful. - Trigger quality: frontmatter
descriptionstates what the skill does and when to use it. - Scope boundaries: each skill has one clear job and avoids becoming a catch-all workflow.
- Progressive disclosure: long guidance lives in
references/; deterministic work lives inscripts/. - Verification: the skill defines commands, evidence, output checks, or review gates.
- Output contract: generated artifacts have stable fields, sections, paths, or acceptance criteria.
- Discoverability: README, help commands, manifests, and platform docs stay synchronized.
- Duplication governance: overlaps are classified as duplicate, strong overlap, complement, or pipeline dependency.
- Safety: side effects, secrets, external services, git operations, publishing, and file writes are bounded.
- Portability: the repo can work across intended harnesses such as Codex, Claude, Cursor, or Agents.
Use this rating:
90-100: benchmark quality
80-89: production ready with minor improvements
70-79: usable but governance gaps remain
60-69: fragile; prioritize cleanup before expansion
<60: do not treat as a high-quality skills repository yetReport Format
Return the audit in Chinese by default:
## 总评
- 总分:
- 等级:
- 主要判断:
## P0 必须修
- [路径] 问题:影响:建议:
## P1 应该修
- ...
## P2 可优化
- ...
## 重复/重叠 Skill
| Skill A | Skill B | 关系 | 建议 |
## 文档与注册面缺口
- README:
- manifest:
- sync:
- 平台文档:
## 30/60/90 分钟迭代计划
- 30 分钟:
- 60 分钟:
- 90 分钟:Resources
scripts/audit_skills_repo.py: deterministic structural audit forskills/repos.references/best-practices.md: detailed rubric derived from strong public skill repositories and local governance patterns.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

