template
Skill1 starsChanged 4 months agoMITView on GitHub
Copy this whole folder to start a new skill: cp -r template skills/<your-skill-name> This template encodes the repo's quality bar so a new skill starts correct-by-construction. Three things make a skill good here: Structure follows PROGRESSIVE DISCLOSURE — load context only as needed: - Layer 1: the frontmatter below (name + description). ALWAYS preloaded. - Layer 2: this SKILL.md body. Loaded when the skill triggers. Keep it lean. - Layer 3: references/*.md. Loaded only when the body tells Claude to…
12 other projects use a version of this.
Adopted by
Most recently taken up by
- davisbuilds/dojotheir own version, first seen Nov 1, 2026
- aihu-project/aihutheir own version, first seen Oct 20, 2026
- o-az/skillstheir own version, first seen Oct 13, 2026
- saqqdy/skillstheir own version, first seen Oct 13, 2026
- G1Joshi/Agent-Skillstheir own version, first seen Oct 1, 2026
- sgomez-dev/claude-skillstheir own version, first seen Sep 30, 2026
Other versions12 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 12
- G1Joshi/Agent-Skills★ 15same name
- amurthygithub/Audit-skills★ 9same name
- aihu-project/aihu★ 4same name
- qhkm/zeptoclaw-skills★ 4same name
- o-az/skills★ 1same name
- saqqdy/skills★ 1same name
- sgomez-dev/claude-skills★ 1same name
- davisbuilds/dojo★ 0same name
- Shugar86/my-agent★ 0same name
- yuka-with-data/skills★ 0same name
2 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 repository21 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- ai-commentsSkill
- ai-draftSkill
- ai-nextSkill
- ai-postSkill
- call-traceSkill
- draft-prSkill
- dto-mappingSkill
- handoffSkill
- hum-commentsSkill
- jira-dcSkill
- meeting-summarySkill
- plan-from-issueSkill
- plan-nextSkill
- plan-reviewSkill
- plan-saveSkill
- pr-commentsSkill
- pr-statsSkill
- review-sessionSkill
- skill-builderSkill
- validator-generatorSkill
The file
We split this skill into the rules it gives an agent
4 rules. 0 are common, 4 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 6 edited versions we read keep these.
4
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 92
If the skill produces structured output, show the EXACT shape in a fenced block — verdict/result first.
- Distinctive: line 126
If this skill needs a big catalog, signature reference, or worked examples, DO NOT inline them here.
- Distinctive: line 127
DO NOT inline them here. Put them in references/ and point to them so they load only when needed, e.g.:
- Distinctive: line 132
Delete references/ entirely if the skill doesn't need it.
“Common” is measured: kept by at least 6 in 10 of the 6 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 TEMPLATE — delete this comment block once you've filled things in.
Copy this whole folder to start a new skill:
cp -r template skills/<your-skill-name>
This template encodes the repo's quality bar so a new skill starts
correct-by-construction. Three things make a skill good here:
1. ONE JOB. A single, clear purpose. If you need "and", split it.
2. DECISION RULES. Hard caps and explicit "do X / never do Y" — not vibes.
3. CONVERGENCE. The skill should reduce churn, not manufacture make-work.
Structure follows PROGRESSIVE DISCLOSURE — load context only as needed:
- Layer 1: the frontmatter below (name + description). ALWAYS preloaded.
- Layer 2: this SKILL.md body. Loaded when the skill triggers. Keep it lean.
- Layer 3: references/*.md. Loaded only when the body tells Claude to read them.
Anything <IN_ANGLE_BRACKETS> is a placeholder for you to fill in. Search the
file for "<" before you ship — no angle brackets should survive.
============================================================================
-->name: <skill-name>
^ lowercase-kebab-case, matches the folder name. This is part of layer-1
metadata and helps Claude decide when to trigger the skill.
version: 0.1.0
description: |
<One or two sentences: WHAT this skill does AND WHEN Claude should use it.>
<The description is the ONLY thing preloaded for every installed skill, and
it is what Claude reads to decide whether to trigger this skill. Lead with
the trigger condition. Be concrete. Avoid "helps with" / "assists" filler.>
allowed-tools:
Minimal set ONLY — list the exact tools this skill needs, nothing more.
A read-only skill should never list Write/Edit/Bash. Delete this comment
and the unused entries below.
- Read
- Grep
- Glob
- Write
- Edit
- Bash
- AskUserQuestion
- Task
<Skill Name>: <one-line statement of the single job>
<!--
Opening frame: orient Claude in 2-4 sentences. State the situation it's in
when this skill fires, the single job, and — if there's a classic failure mode
for this kind of task — name it up front (the plan-review skill opens by naming
the context-gap that causes bad reviews). Delete this comment.
--><You are about to .... Your job is to .... Do X, then STOP — do not Y.>
When to use this skill
<!-- Crisp triggers. Mirror the description but in list form. -->- <Trigger 1>
- <Trigger 2>
When NOT to use this skill
<!-- Just as important. Prevents misfires and scope creep. -->- <Anti-trigger 1 — "use
<other-skill>instead"> - <Anti-trigger 2>
Steps
<!--
Numbered, imperative, in order. Each step is one action. Include EXPLICIT
stop conditions ("if X, ask the user and stop"). Never let the model guess
when it's blocked — route to AskUserQuestion. End with an explicit "Stop."
--><First action.><Detail. What to read first, what to look for.><Next action.><Detail.>- <...>
N. Stop. <State exactly what "done" is and that the skill does not continue past it.>
Output format
<!--
If the skill produces structured output, show the EXACT shape in a fenced
block — verdict/result first. If it edits files instead, replace this section
with the file-format spec. If it produces no fixed output, delete this section.
--><Show the literal output structure here. Put the most important line first.>Rules
<!--
This is where "decision rules over vibes" lives. Use bold lead-ins. Prefer
hard caps and "NEVER / Do NOT" over soft advice. If the skill makes judgment
calls, give it a "what to flag / what NOT to flag" pair so the boundary is
explicit. Keep universal opinions framed as configurable, not law.
-->What to do
- <Rule.> <Why, briefly.>
- <Hard cap, if any.> <e.g. "Max 2 X. Hard cap — pick the two most important.">
What NOT to do
- <Anti-rule.> <The tempting-but-wrong behavior, and what to do instead.>
- NEVER
<the cardinal sin for this skill>.
Format discipline
- <No preamble / no recap / verdict-first / stop when done — whatever keeps the output tight.>
<!--
LAYER 3 — PROGRESSIVE DISCLOSURE
If this skill needs a big catalog, signature reference, or worked examples,
DO NOT inline them here. Put them in references/ and point to them so they
load only when needed, e.g.:
For the full <thing> catalog, read `references/<thing>-catalog.md`.
Delete references/ entirely if the skill doesn't need it. Likewise keep evals/
and fill it in — every skill in this repo ships a way to prove it works.
-->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.

