skill-review
Skill0 starsChanged 17 days agoMITView on GitHub
Reviews installed agent skills for unused, misfiring or outdated ones, and turns the user's repeated corrections into skill fixes. Use for a periodic skill check-up, after a new model release, or when skills seem to be ignored. Step 4 of `skill-authoring`.
25 other projects use a version of this.
Adopted by
Most recently taken up by
- snuffish/dotfilestheir own version, first seen Nov 8, 2026
- IT-HUSET/andthentheir own version, first seen Nov 8, 2026
- peterpme/skillstheir own version, first seen Nov 8, 2026
- pillumina/ascend-sleuththeir own version, first seen Nov 2, 2026
- AbdouMoumen/claude-skills-plugintheir own version, first seen Nov 1, 2026
- AlbertPhan88/claude-skillstheir own version, first seen Oct 20, 2026
Used with
Projects with it also use sigma, skill-forge and 2 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
sanyuan0704Skillsigma
Personalized 1-on-1 AI tutor using Bloom's 2-Sigma mastery learning. Guides users through any topic with Socratic questioning, adaptive pacing, and rich visual output (HTML dashboards, Excalidraw concept maps, generated images). Use when user wants to learn something, study a topic, understand a concept, requests tutoring, says 'teach me', 'I want to learn', 'explain X to me step by step', 'help me understand', or invokes /sigma. Triggers on: learn, study, teach, tutor, understand, master, explain step by step.
Also in 3 projects that use this
WilliamSaysXSkillskill-forge
Automated skill creation workshop with intelligent source detection, smart path management, and end-to-end workflow automation. This skill should be used when users want to create a new skill or convert external resources (GitHub repositories, online documentation, or local directories) into a skill. Automatically fetches, organizes, and packages skills with proactive cleanup management.
Also in 3 projects that use this
SentrySkillskill-creator
Create new agent skills following the Agent Skills specification. Use when asked to "create a skill", "add a new skill", "write a skill", "make a skill", "build a skill", or scaffold a new skill with SKILL.md. Guides through requirements, writing, registration, and verification.
Also in 4 projects that use this
GoogleSkillgrill-me
A relentless interview to sharpen a plan or design.
Also in 3 projects that use this
Other versions25 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 25
- sanyuan0704/code-review-expert★ 3.9ksame name
- sanyuan0704/sanyuan-skills★ 3.9ksame name
- QoderAI/better-harness★ 2.4ksame name
- fagemx/gstack-game★ 68same name
- joelhooks/joelclaw★ 65same name
- coupler-io/skills★ 39same name
- nimadorostkar/Claude-Skills-collection★ 33same name
- peterpme/skills★ 28same name
- pillumina/ascend-sleuth★ 5same name
- IT-HUSET/andthen★ 4same name
15 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository9 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- friction-miningSkill
- friction-probeSkill
- skill-authoringSkill
- skill-scopingSkill
- skill-testingSkill
- skill-vettingSkill
- skill-writingSkill
The file
We split this skill into the rules it gives an agent
5 rules. 0 are common, 4 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 20 edited versions we read keep these.
4
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 23
Never used, though the user did that kind of task: it under-triggers.
- Distinctive: line 24
Never used, and nobody does that task: a candidate to hide or retire.
- Distinctive: line 29
Read only the sessions where the reviewed skill ran:
- Distinctive: line 42
Never copy secrets or personal data from a transcript into a skill.
- Distinctive: line 65
Run monthly and after each model release.
“Common” is measured: kept by at least 6 in 10 of the 20 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Skill review
Skills decay quietly: descriptions stop matching how the user asks, new models outgrow them, and corrections the user keeps repeating never reach the skill. Review on a schedule, and change nothing without the user's yes.
1. Usage
uv run scripts/usage.py --days 30It reads the local Claude Code and Codex transcripts, read-only, and reports per installed skill the number of sessions that used it, per agent, and the last use, plus installed skills never used. In Claude Code, /skill-doctor adds what each skill costs in context. Read the result like this:
- Never used, though the user did that kind of task: it under-triggers. Run trigger evals with
skill-testing. - Never used, and nobody does that task: a candidate to hide or retire.
- Used often: the skill most worth testing well.
2. Corrections
Read only the sessions where the reviewed skill ran:
uv run scripts/usage.py --sessions SKILL --days 30In each, find the moments after the skill loaded where the user corrected the agent or explained the same thing again. Quote each correction word for word, then route it:
- caused by the skill's instructions or a gap in them: a gotcha and an eval case for that skill (
skill-writing, thenskill-testing); - a rule the skill states and the agent broke: set the skill's kind to discipline and run pressure tests (
skill-testing).
Repeated friction outside any skill's sessions is not this step: load friction-mining for it.
Never copy secrets or personal data from a transcript into a skill.
3. New model release
Re-run each skill's evals with the new model pinned (skill-testing). If the no-skill arm now passes, the skill isn't broken; it is no longer needed. Propose retiring it, starting with capability-uplift skills. If results drop, the skill needs rework.
4. Decide per skill
| Finding | Proposal |
| Works and gets used | keep |
| Under-triggers or false-triggers | fix the description (skill-writing, then trigger evals) |
| Repeated corrections | add gotchas and eval cases (skill-writing) |
| Overlaps another skill, or does two jobs | rescope (skill-scoping, rework mode) |
| Rarely needed but worth keeping | hide it: Claude Code skillOverrides set to name-only or off; disable it in Codex config |
| Outgrown by the model, or dead | retire it: status: deprecated, a CHANGELOG entry, callers updated, removed in the next release |
5. Report
Fill in the report template and send it in one message. Ask for a yes per proposal, then apply only the approved ones through the skill named in the table.
Schedule
Run monthly and after each model release. A recurring run (cron, a scheduled agent task) changes the user's setup: propose it, and set it up only after a yes. An unattended run only writes the report; it never edits a skill.
Gotchas
- Claude Code deletes old transcripts (after 30 days by default), so "never used" means "not in the kept history".
- A skill the user starts by typing
/namemay not appear as a Skill tool call, sousage.pycan undercount skills invoked by hand. - Codex counts are heuristic: they count tool calls that read a skill's SKILL.md, not the skill list Codex puts in its own prompt.
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.