review-skill skill
Evaluates Agent Skills for structural validity, instruction clarity, security risks, and accuracy of technical claims.
- GitHub stars
- 331
- Files
- 2
- Last commit
- Oct 10, 2026
- License
- MIT
- Allowed tools
- —
What does the review-skill skill do?
review-skill evaluates Agent Skills for structural validity, instruction clarity, security risks, and accuracy of technical claims.
Summary generated from the skill's SKILL.md.
About the review-skill skill
Developers who build and maintain Agent Skills can use this tool to audit their documentation and technical claims for accuracy. It performs a deep inspection of SKILL.md files to ensure they follow structural requirements, use correct frontmatter, and provide clear instructions. The process identifies missing inputs, silent failure paths, and potential security risks like credential exposure or destructive command scopes.
When an agent loads this skill, it acts as a rigorous reviewer that tests examples against cited APIs or project sources. It can run code in empty scratch directories using pnpm-workspace.yaml to verify that exit codes, error texts, and counts match what the documentation claims. The tool requires an agent with filesystem and shell access, plus network access for remote sources. It categorizes every issue as an error, warning, or note, providing the exact file path and line number alongside a direct fix.
What an agent can do with it
- 1Verify technical documentation accuracyCheck if the code examples and prose claims in a skill file match the actual behavior of the software version specified.
- 2Audit for security vulnerabilitiesScan skill commands for dangerous behaviors such as unverified downloads or the exposure of sensitive credentials.
Written from the skill's SKILL.md. Every use case is checked against a quote from it.
How to install the review-skill skill
Each command copies the skill folder into the client's skills directory. Restart the client or start a new session to load it.Install the review-skill skill in Plugin
/plugin marketplace add skilld-dev/skilld /plugin install skilld@skilld
Install the review-skill skill in Claude Code
~/.claude/skills/review-skill/SKILL.md
$ npx degit skilld-dev/skilld/skills/review-skill ~/.claude/skills/review-skillInstall the review-skill skill in Claude Code (project)
.claude/skills/review-skill/SKILL.md
$ npx degit skilld-dev/skilld/skills/review-skill .claude/skills/review-skillInstall the review-skill skill in Codex
~/.agents/skills/review-skill/SKILL.md
$ npx degit skilld-dev/skilld/skills/review-skill ~/.agents/skills/review-skillInstall the review-skill skill in Gemini CLI
~/.gemini/skills/review-skill/SKILL.md
$ npx degit skilld-dev/skilld/skills/review-skill ~/.gemini/skills/review-skillInstall the review-skill skill in Claude.ai
npx degit skilld-dev/skilld/skills/review-skill review-skill && zip -r review-skill.zip review-skill # claude.ai → Customize → Skills → + → Create skill → Upload a skill → review-skill.zip
Automated checks
Ran Nov 14, 2026Full report →Not verified yet. A skill needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachableNot checked
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit 2026-10-10github.com/skilld-dev/skilld/commits ↗Pass
- No download-and-run or credential theft in scriptsNo scripts in the skill folderPass
- No hidden instructions in SKILL.mdNo override phrases, safeguard switches, credential requests or hidden charactersgithub.com/skilld-dev/skilld/blob/HEAD/skills/review-skill/S ↗Pass
- Pre-approved tools are scopedNo pre-approved tools; the agent asks before using onegithub.com/skilld-dev/skilld/blob/HEAD/skills/review-skill/S ↗Pass
- SKILL.md follows the Agent Skills formatthe description does not say when to use the skillgithub.com/skilld-dev/skilld/blob/HEAD/skills/review-skill/S ↗Finding
SKILL.md
View on GitHubReview a Skill
Review the supplied Skill as an Agent would use it.
Checks
- Confirm
SKILL.mdexists and its parent directory matches its name. - Confirm frontmatter uses supported fields and valid values.
- Confirm the description says what the Skill does and when to use it, in the third person, with the terms a user types.
- Follow every linked reference and script.
- Report missing or broken links.
- Reject symbolic links, special files, and paths that leave the Skill directory.
- Check instructions for missing inputs, unclear outcomes, and silent failure paths.
- Check commands for destructive scope, credential exposure, and unverified downloads.
- Check examples against the cited API or project source. If a runtime is available, run them and report each result that differs from the claim.
An example is each code block and each checkable prose claim: a default, a list, a count, error text, an exit code, or "X happens when Y".
Test the exact version the Skill names, never the repository head. Use the toolchain its users run, in each mode the claim covers, such as SSR, dev, and build.
Install it in an empty scratch directory with its own
pnpm-workspace.yaml. Prefer a local fixture to a live site. Never use personal credentials. For cloud-bound code, typecheck it and run it against in-memory stand-ins. Confirm a failing claim with a second probe before you report it. Mark each example executed, read from source, or unverified. - Find repeated prose and material that belongs in a reference.
- Find text the reader already knows: domain or framework explanations, generic debug advice, changelog paraphrase, and internals the reader cannot act on.
- For a package Skill, confirm the body names the package version it was tested against, once, in the first paragraph. Flag other version literals.
- Confirm the body covers each export, option, or symptom the description promises.
Rank each finding:
error: following the Skill gives a wrong result, a failure, or a security gap.warning: an Agent would act wrongly on the claim, a failure stays silent, or a needed input is missing.note: precision, repetition, structure, or token cost.
Give the exact path, the line, and a direct fix. Do not rewrite the Skill unless the request asks for changes.
For a direct run, present the findings to the user. The user decides whether to apply them.
Files in this skill
2review-skill skill questions
How do I install the review-skill skill in Claude Code?
Run npx degit skilld-dev/skilld/skills/review-skill ~/.claude/skills/review-skill. Claude Code loads skills from ~/.claude/skills (all projects) or .claude/skills (one project) and uses a skill when its description matches the task.
Which clients can use the review-skill skill?
Any client that reads the Agent Skills format: Claude Code, Codex, Gemini CLI and claude.ai (upload as a ZIP). Scripts in the skill run only where the client allows code execution.
What specific aspects of a skill does this tool check?
It verifies the SKILL.md structure, frontmatter validity, description clarity, link integrity, instruction completeness, and command safety.
How does the tool handle code examples and API claims?
It tests examples against cited sources or runtimes using specific versions, installs them in scratch directories, and marks them as executed, read, or unverified.