agentleFS
spencerpaulyspencerpauly/skills-repo/skills/suggesting-skills/SKILL.md

suggesting-skills

Skill25 starsChanged 7 months agoMITView on GitHub

Look at what the user keeps re-explaining to the agent and propose new SKILL.md files that would replace those repeated explanations. Use when the user says "what skills should I have", "suggest skills", "audit my skills", or when you notice you're being asked the same kind of task three times in a row.

3 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions3 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 3

    1. witi-projects/Skills★ 2same name
    2. ZeroTokenClaw/agents-skills★ 1same name
    3. matarpeles/skills-registry-demo★ 0same name

    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 repository5 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 4 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    After a long session, before the user closes the chat, offer to extract any reusable playbook from the work just done.

    line 15
  2. Very specific:

    Inventory what's already there: Read every SKILL.md in the repo.

    line 19
  3. Distinctive:

    Ask the user which 1–2 to actually write: Don't auto-generate all of them.

    line 37
  4. Distinctive:

    Don't suggest a skill for something the agent already does well by default.

    line 61
  5. Distinctive:

    Don't propose 20 skills. The repo gets noisy and the agent's skill-selection routing degrades.

    line 62
  6. Very specific:

    Don't write the SKILL.md without the user's go-ahead.

    line 63

“Common” is measured: kept by at least 6 in 10 of the 3 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

Suggesting Skills

Find the gaps in the user's skill library. Skills are most valuable when they capture the things the user would otherwise have to explain to the agent every time.

When to use

  • The user explicitly asks "what skills am I missing", "what should I add", "audit my skills".
  • You notice the same kind of request happening repeatedly in a session — that's a skill waiting to be written.
  • After a long session, before the user closes the chat, offer to extract any reusable playbook from the work just done.

Steps

  1. Inventory what's already there. Read every SKILL.md in the repo. Note for each: name, the trigger phrase in the description, and the one-line goal. If two existing skills overlap, flag the duplication.
  2. Look for the pattern of "I keep telling you to…". Strong candidates for new skills are:
    • Repeated workflow — the user has asked for the same multi-step task at least twice.
    • Repeated correction — the user has had to fix the same kind of mistake the agent makes (formatting, voice, missing a step).
    • Tribal knowledge — facts about this user's stack, team, or product that aren't in the codebase but the user keeps having to restate.
    • External tool wiring — the user has an MCP, CLI, or API the agent could be using but isn't.
  3. Propose 3–5 candidate skills, each as a one-paragraph pitch:
    Name:        kebab-case-name
    Trigger:     when the user says "..." or asks to "..."
    What it does: one sentence
    Why it's worth a skill: the repeated thing it replaces
  4. Rank them. Sort by frequency × annoyance: how often does the situation come up, and how painful is it when it's not automated?
  5. Ask the user which 1–2 to actually write. Don't auto-generate all of them. Skills that aren't trusted get ignored, which is worse than not having them.
  6. For each chosen skill, scaffold a SKILL.md in skills/<name>/ with:
    • Frontmatter (name, description with trigger phrases, license).
    • "When to use" — the trigger conditions, in plain English.
    • "Steps" — numbered, imperative.
    • "Don't" — the failure modes you've seen.

Output format

## Existing skills (N)
- skill-name — what it does, when it fires.

## Suggested additions
1. **<name>** — <pitch> _(seen X times this session)_
2. ...

## Candidate skill files
<scaffolded SKILL.md(s) for the chosen ones>

Don't

  • Don't suggest a skill for something the agent already does well by default. Skills are for the cases where the un-prompted behavior is wrong.
  • Don't propose 20 skills. The repo gets noisy and the agent's skill-selection routing degrades. 5–15 well-trusted skills > 50 mediocre ones.
  • Don't write the SKILL.md without the user's go-ahead. Half the value of a skill is that the user trusts it; that requires their hand on the design.

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.