agentleFS
tejastn10tejastn10/workbench/skills/meta/write-skill/SKILL.md

write-skill

Skill1 starsChanged 2 months agoMITView on GitHub

Author a new skill for this repo in the house style — YAML frontmatter with a trigger-rich description, terse mechanism-focused body, a "what to let slide" section for review skills, an anti-patterns list, grounded in real conventions not generic advice. Use when adding a skill to workbench, when asked to "write a skill for X", or to fix an existing one that's too vague.

25 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use write-docs, ship and 6 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.
  • SentrySkill

    write-docs

    Create or update documentation pages for the Abacus docs site. Use when adding new documentation, updating existing pages, or ensuring docs stay in sync with code changes. Ensures pages follow Starlight patterns and build correctly.

    Also in 3 projects that use this

  • AWSSkill

    ship

    Finalize dev work on an issue branch — update docs, push, and open a PR.

    Also in 4 projects that use this

  • Google CloudSkill

    triage

    Gather context on a change or bug and plan the implementation.

    Also in 4 projects that use this

  • SentrySkill

    review

    Fixture review skill.

    Also in 5 projects that use this

  • AWSSkill

    retro

    Conduct a retrospective on a coding session.

    Also in 3 projects that use this

  • Hugging FaceSkill

    implement

    Make tests pass. Invoke after /write-tests produces failing tests.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 23 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep it one sentence-ish; front-load what it does.

    line 24
  2. Very specific:

    Add the dir path (./skills/<category>/<name>) to the skills array in .claude-plugin/plugin.json — the plugin manifest is explicit, not …

    line 49
  3. Very specific:

    Add a row to skills/README.md and the category README.md.

    line 51
  4. Very specific:

    If it's a template-filler, the template goes in docs/templates/ and its purpose row in docs/templates/README.md.

    line 52

“Common” is measured: kept by at least 6 in 10 of the 23 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Write a skill

Adapted in spirit from mattpocock/skills writing-for-agents. A skill is a packaged instruction set an agent loads instead of its default approach — so it must be specific enough to change behaviour.

Frontmatter

---
name: <kebab-case>            # matches the filename; unique across the repo
description: <what it does — one clause> · <the distinctive detail> · Use when <triggers>
---
  • The description is the router — the agent decides to load the skill from it. Pack it with the phrases someone would actually use ("review this PR", "cut a release", "why is X slow"). Name the stack / tool if it's specific.
  • Keep it one sentence-ish; front-load what it does.

Body

  • Terse. Same register as pr-review/nestjs-backend-pr-review: short sentences, name the mechanism, no throat-clearing.
  • Grounded, not generic. "Use SET k v EX 300, not SET then EXPIRE" beats "handle TTLs carefully". Pull real rules from the codebase, the review corpus, the actual tooling. If you're writing best-practice platitudes, stop.
  • Structure to the task:
    • Review skills: ranked "what to flag" (with the reason), a "what to let slide" section (as important), how to structure the output, then order-of-operations.
    • Process skills: the steps in order, the rules, the output.
  • Reference sibling skills as `category/name` (bare, no path) — the agent invokes by name. The site turns Adapted from [source] at the start of a line into a badge.
  • End with ## Anti-patterns — what the agent must not do, as ❌ bullets.
  • ~40–120 lines. Longer means it's trying to be a textbook.

Placement

  • skills/<category>/<name>/SKILL.md — one directory per skill (Claude Code plugin format). New category → just make the folder + a README.md; the site auto-discovers it and derives the label from the slug.
  • Add the dir path (./skills/<category>/<name>) to the skills array in .claude-plugin/plugin.json — the plugin manifest is explicit, not auto-discovered.
  • Add a row to skills/README.md and the category README.md.
  • If it's a template-filler, the template goes in docs/templates/ and its purpose row in docs/templates/README.md.

Anti-patterns

  • ❌ A description that doesn't say when to use the skill.
  • ❌ Generic best-practice advice with no line-level specifics.
  • ❌ A review skill with no "what to let slide" section.
  • ❌ No anti-patterns list.
  • ❌ Restating what the tool's own docs say instead of the house opinion.
  • ❌ 300 lines.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.