agentleFS
jcurbelojcurbelo/agent-seed/.claude/skills/write-a-skill/SKILL.md

write-a-skill

Skill8 starsChanged 8 months agoMITView on GitHub

Create new skills for this agent-seed install with proper frontmatter, progressive disclosure, and conventions (size limits, index updates, voice integration). Use when the user wants to create, write, or build a new skill in `.claude/skills/`.

126 other projects use a version of this, 93 of them edited.

Adopted by

Most recently taken up by

Used with

Projects with it also use zoom-out, to-issues and 9 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.
  • victorgarciaesgiSkill

    zoom-out

    Tell the agent to zoom out and give broader context or a higher-level perspective. Use when you're unfamiliar with a section of code or need to understand how it fits into the bigger picture.

    Also in 48 projects that use this

  • GoogleSkill

    to-issues

    Break a plan, spec, or PRD into independently-grabbable issues on the project issue tracker using tracer-bullet vertical slices.

    Also in 43 projects that use this

  • kennethkeimSkill

    ubiquitous-language

    Extract a DDD-style ubiquitous language glossary from the current conversation, flagging ambiguities and proposing canonical terms. Saves to UBIQUITOUS_LANGUAGE.md. Use when user wants to define domain terms, build a glossary, harden terminology, create a ubiquitous language, or mentions "domain model" or "DDD".

    Also in 20 projects that use this

  • GoogleSkill

    to-prd

    Turn the current conversation into a PRD and publish it to the project issue tracker — no interview, just synthesis of what you've already discussed.

    Also in 42 projects that use this

  • HK-hubSkill

    prd-to-plan

    Turn a PRD into a multi-phase implementation plan using tracer-bullet vertical slices, saved as a local Markdown file in ./plans/. Use when user wants to break down a PRD, create an implementation plan, plan phases from a PRD, or mentions "tracer bullets".

    Also in 7 projects that use this

  • semtexfromuaSkill

    how-to

    Курс typing-уроків по існуючому проекту — «як написати його з нуля»: аналіз → план у .practicum/course.md → покрокова генерація уроків. Використовуй для команди /vibe-typing:how-to.

    Also in 3 projects that use this

  • +2 more from google

The file

We split this skill into the rules it gives an agent

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

  • 1

    Common

    Most of the 122 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use this when the user has a recurring task that deserves its own slash command.

    line 7
  2. Common:

    Should any section be more / less detailed?

    line 26
  3. Very specific:

    Add the new skill to the ## Skills list in README.md

    line 29
  4. Very specific:

    If a .claude/skills/index.md exists, add an entry there too

    line 30
  5. Very specific:

    Add scripts/ utilities when:

    line 84

“Common” is measured: kept by at least 6 in 10 of the 122 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 - Skill Authoring

Use this when the user has a recurring task that deserves its own slash command. Often invoked from /learn when a gap is detected.

Process

  1. Gather requirements — ask the user:
    • What task does the skill cover? What triggers it?
    • Single-step or multi-step (does it dispatch agents)?
    • Does it produce output? Where does the output land?
    • Does it need bundled scripts, or just instructions?
    • Reference materials to include?
  2. Draft the skill — create:
    • SKILL.md with concise instructions (target under 150 lines, hard cap at 250 per repo policy)
    • Additional reference files if content exceeds limit (link from SKILL.md)
    • Utility scripts in scripts/ if deterministic operations are needed
  3. Review with the user — present the draft and ask:
    • Does this cover the use case?
    • Anything missing or unclear?
    • Should any section be more / less detailed?
  4. Wire it up:
    • Add the new skill to the ## Skills list in README.md
    • If a .claude/skills/index.md exists, add an entry there too
    • Validate with scripts/check-markdown-policy.sh

Skill Structure

.claude/skills/skill-name/
├── SKILL.md           # Main instructions (required)
├── REFERENCE.md       # Detailed docs (only if SKILL.md > 150 lines)
├── EXAMPLES.md        # Usage examples (only if needed)
└── scripts/           # Utility scripts (only if deterministic ops needed)
    └── helper.sh

SKILL.md Template

---
description: Brief description of capability. Use when [specific user triggers — their own phrasing].
---

# /skill-name - Short Title

## Input

What the user provides when invoking.

## Step 1: [first action]

[Concrete instructions. If dispatching agents, launch independent ones in parallel.]

## Step 2: [next action]

...

## Output

What the skill produces (file path, draft, summary).

Conventions (agent-seed)

These differ from generic skills — apply them:

  • Description — must include "Use when [specific triggers]". This is the only thing the model sees when deciding to load the skill. Use the user's actual phrasing.
  • Model tiering — if the skill dispatches agents, declare model in the agent definition (haiku for simple lookups, sonnet for analysis / coding, opus for orchestration / complex reasoning). See CLAUDE.md § Model Tiering.
  • Parallel dispatch — when multiple agents are independent, launch them in a single message with multiple tool calls. Never sequentially.
  • Output location — drafts and notes go inside the repo (todos/YYYY-MM-DD/, handoffs/, or a skill-specific dir). Never /tmp.
  • File size limits — skills under 150 lines (link reference files if larger). Enforced by scripts/check-markdown-policy.sh.
  • Voice — anything user-facing (drafted messages, PRs, emails) follows context/voice/index.md and context/voice/patterns.md. Skills should reference those files, not duplicate the voice rules.
  • Index pattern — if the skill produces files in a new folder, follow the index.md + detail pages convention.

When to Add Scripts

Add scripts/ utilities when:

  • Operation is deterministic (validation, formatting, URL building)
  • Same code would otherwise be regenerated every invocation
  • Errors need explicit handling

Scripts save tokens and improve reliability vs generated code.

Description Writing

The description is the only thing the model sees when deciding which skill to load.

  • Max 1024 chars
  • Third person
  • First sentence: what it does
  • Second sentence: "Use when [specific triggers]"

Good:

Compact the current conversation into a handoff document so a fresh agent can continue the work. Use when the conversation is approaching context limits or you want to hand off to a new session.

Bad:

Helps with handoffs.

Review Checklist

After drafting, verify:

  • Description includes specific triggers ("Use when...")
  • SKILL.md under 150 lines (link reference files if larger)
  • No time-sensitive info (dates, "currently", "as of...")
  • Consistent terminology with the rest of the repo
  • Concrete examples included
  • References to context/, memory/, and .claude/rules/ where relevant — don't duplicate
  • Output paths are inside the repo (never /tmp)
  • README.md ## Skills list updated
  • scripts/check-markdown-policy.sh passes

Adapted from mattpocock/skills/productivity/write-a-skill. See memory/external-skill-sources.md.

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.