agent-skill-reviewer
Skill87 starsChanged 6 months agoApache-2.0View on GitHub
Review Agent Skill directories and SKILL.md files against best practices. Use this skill when the user wants to review, validate, or check an Agent Skill implementation.
In this repository21 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- code-readerSkill
- dir-organizerSkill
- doc-reviewerSkill
- drawio-designerSkill
- editorial-card-designerSkill
- humanizer-zhSkill
- md-link-checkerSkill
- md-summarizerSkill
- md-translatorSkill
- ontologySkill
- openspec-assistantSkill
- pptx-editorSkill
- pptx-readerSkill
- project-analyzerSkill
- reference-organizerSkill
- self-reviewSkill
- tech-outline-plannerSkill
- update-submitterSkill
- web-content-downloaderSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
13 rules. About 0 are common, 4 are distinctive, and 9 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
9
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 10
When the user asks you to review an Agent Skill, follow these steps to ensure the skill complies with the best practices.
- Distinctive: line 18
Check if the project follows the production-grade directory structure:
- Very specific: line 20
Directory Name: Must use kebab-case.
- Very specific: line 21
Core File: Ensure the core instruction file is named exactly SKILL.md (uppercase).
- Very specific: line 22
Separation of Concerns: Look for proper subdirectories (e.g., scripts/ for executable code, references/ for docs, assets/ for static files).
- Very specific: line 26
Role-based Naming: Check if the skill name uses a Noun/Doer format instead of a Verb/Action format (e.g., agent-skill-reviewer instead of …
- Very specific: line 30
Read the YAML frontmatter of SKILL.md.
- Very specific: line 31
Name: Verify that the name field is present and matches the directory name.
- Very specific: line 32
Description Formula: Verify that the description follows the golden formula: [Function] + [Trigger Scenario] + [Keywords].
- Distinctive: line 36
Clarity: Check if the instructions are clear, step-by-step, and easy for an LLM to follow.
- Distinctive: line 37
State & Concurrency: Ensure the skill does not attempt to handle highly concurrent parallel tasks within a single conversation thread, as …
- Very specific: line 38
Offloading: Ensure no complex logic is hardcoded in Markdown if it can be offloaded to an external tool or script (e.g., in scripts/).
- Very specific: line 39
Progressive Disclosure: Verify that detailed supplementary knowledge is placed in references/ rather than cluttering the main SKILL.md.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
Agent Skill Reviewer
Instructions
When the user asks you to review an Agent Skill, follow these steps to ensure the skill complies with the best practices.
1. Identify the Target
Determine the directory path of the Agent Skill to be reviewed. Read the directory structure and the content of the SKILL.md file.
2. Analyze Directory Structure
Check if the project follows the production-grade directory structure:
- Directory Name: Must use
kebab-case. - Core File: Ensure the core instruction file is named exactly
SKILL.md(uppercase). - Separation of Concerns: Look for proper subdirectories (e.g.,
scripts/for executable code,references/for docs,assets/for static files).
3. Review Naming Conventions
- Role-based Naming: Check if the skill name uses a Noun/Doer format instead of a Verb/Action format (e.g.,
agent-skill-reviewerinstead ofagent-skill-review).
4. Review Frontmatter (Metadata Layer)
- Read the YAML frontmatter of
SKILL.md. - Name: Verify that the
namefield is present and matches the directory name. - Description Formula: Verify that the
descriptionfollows the golden formula:[Function] + [Trigger Scenario] + [Keywords]. It must be specific enough for the LLM to reason about when to trigger the skill.
5. Review Instructions (Core Logic Layer)
- Clarity: Check if the instructions are clear, step-by-step, and easy for an LLM to follow.
- State & Concurrency: Ensure the skill does not attempt to handle highly concurrent parallel tasks within a single conversation thread, as skills modify the context state.
- Offloading: Ensure no complex logic is hardcoded in Markdown if it can be offloaded to an external tool or script (e.g., in
scripts/). - Progressive Disclosure: Verify that detailed supplementary knowledge is placed in
references/rather than cluttering the mainSKILL.md.
6. Report Findings
Present a structured review report to the user using the following format:
- Directory Structure & Naming: Pass/Fail with details.
- Frontmatter: Pass/Fail with details.
- Instructions: Pass/Fail with details.
- Recommendations: Provide actionable recommendations for any violations of best practices.
References
references/best-practices.md— Detailed best-practice checklist for agent skill directory structure, naming conventions, trigger descriptions, progressive disclosure, and language conventions. Load this when you need the full specification behind the review criteria.
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.

