agentleFS
qyndexqyndex/claude-init/.claude/skills/skill-creator/SKILL.md

skill-creator

Skill2 starsChanged 5 months agoMITView on GitHub

Use when a recurring task pattern emerges that deserves its own skill. Generates a fresh SKILL.md under .claude/memory.proposed/skills/<slug>/ (NOT activated until /dream-review --approve-skill). Triggers — manual (/create-skill), instinct promotion (--as-skill on a mature instinct), or task-signature-detector (3+ Bash sequence repeated ≥5× across ≥2 sessions). Round 9 A.

  • Pipes a download into a shell
  • Deletes or force-pushes

189 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use taskflow-inbox-triage, mermaid-diagram-generator and 10 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.
  • AtomicBot-aiSkill

    taskflow-inbox-triage

    Also in 6 projects that use this

  • shalombSkill

    mermaid-diagram-generator

    Guide on creating and editing Mermaid charts safely. Use when asked to draw a diagram, flowchart, or architecture chart. Triggers on: 'mermaid', 'diagram', 'flowchart', 'visualize'.

    Also in 5 projects that use this

  • naman10parikhSkill

    ecosystem-reference

    Quick-access directory of Claude Code ecosystem repos, skills, plugins, MCPs, and patterns. Use when improving the harness.

    Also in 5 projects that use this

  • naman10parikhSkill

    fractal-delegation

    Also in 5 projects that use this

  • Liang-MHSkill

    stock_analysis

    获取 A 股、基金和港股的实时行情和技术分析数据

    Also in 5 projects that use this

  • nicepkgSkill

    community-led-growth

    Expert in community-led growth (CLG) - leveraging user communities to drive acquisition, retention, and expansion. Covers building developer communities, user groups, ambassador programs, and turning customers into advocates. Knows the difference between community as a feature vs community as a growth engine, and how to measure community ROI. Use when "community-led, CLG, ambassador program, champion program, user community, developer community, word of mouth, user group, " mentioned.

    Also in 5 projects that use this

  • +6 more from naman10parikh

The file

We split this skill into the rules it gives an agent

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

  • 0

    Common

    Most of the 185 edited versions we read keep these.

  • 0

    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. Very specific:

    NEVER write directly to .claude/skills/: Always .claude/memory.proposed/skills/.

    line 83
  2. Very specific:

    NO file created under .claude/skills/ (that's the approval step)

    line 114

“Common” is measured: kept by at least 6 in 10 of the 185 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

Skill Creator

You generate new skills from observed patterns. Your output is a proposal, never an active skill — a human approves via /dream-review --approve-skill <slug> before activation.

Inputs

  • description (required) — one-line "what this skill does, when to invoke"
  • slug (optional) — kebab-case; auto-derived from description if absent
  • examples[] (optional) — paths or transcript snippets showing the recurring pattern
  • triggers[] (optional) — phrases that should activate the skill via skill-router.sh
  • model (optional) — inherit | sonnet | opus | haiku; default inherit

Output

Write to .claude/memory.proposed/skills/<slug>/:

SKILL.md         # frontmatter + body (matches existing skill conventions)
EXAMPLES.md      # 3-5 canonical worked invocations
triggers.yml     # phrase patterns for skill-router.sh (one per line)
provenance.md    # who/what generated this; observation source; reviewer

SKILL.md frontmatter contract

Required:

  • name: <slug>
  • description: <one-line, lead with "Use when…", ≤ 1536 chars>
  • when_to_use: <concrete signals that trigger> (Round 9 F requirement)
  • model: inherit | sonnet | opus | haiku

Optional:

  • disable-model-invocation: false (default; set true for operator-only skills)
  • triggers: [<phrase1>, <phrase2>]
  • allowed-tools: <subset>
  • permissionMode: default | plan | acceptEdits

SKILL.md body sections (in order)

  1. # <Skill Name> — title (Title Case)
  2. Mandate — 2-3 sentences: what this skill enforces / produces
  3. Inputs — what the operator/agent provides
  4. Procedure — numbered steps; cite tools/commands; ≤ 12 steps; if more, decompose
  5. Hard rules — invariants the skill enforces; bulleted
  6. Examples — 1-3 invocations linking to EXAMPLES.md
  7. Done means — the exit criteria

Validations (run BEFORE writing the file)

CheckAction on fail
Frontmatter has name, description, when_to_use, modelBlock; re-prompt
Description starts with "Use when…"Block; re-prompt
Description ≤ 1536 charsBlock; truncate or re-prompt
Body ≤ 400 LOC (CLAUDE.md §XIV)Block; decompose into multiple skills
gitleaks detect on bodyBlock if secret detected
No destructive ops in body: rm -rf, `curlsh, eval, --no-verify, base64sh`Block; replace with safer pattern
slug is kebab-case, lowercase, ≤ 32 charsBlock; auto-fix
.claude/skills/<slug>/ does NOT already existBlock; suggest different slug
Per-day creation count ≤ 3Block; rate-limit

Procedure

  1. Receive trigger — manual /create-skill <slug> "<description>", instinct promotion, or task-signature-detector candidate.
  2. Read evidence — if examples[] provided, read each; else look at observations.jsonl + bash.log for the recurring signature.
  3. Draft SKILL.md — fill the body sections above. Body lines ≤ 400.
  4. Draft EXAMPLES.md — 3 invocations showing input → process → output.
  5. Draft triggers.yml — 3-8 phrase patterns for skill-router.sh hint mapping.
  6. Validate — run every check in the table above. Refuse on any failure.
  7. Write to .claude/memory.proposed/skills/<slug>/ — atomic, never overwrite an existing proposal.
  8. Emit NEXUS YAML handoff — status: completed | blocked; artifacts_created: [...]; followup_tasks: [{summary: "/dream-review --approve-skill <slug>"}].

Hard rules

  • NEVER write directly to .claude/skills/. Always .claude/memory.proposed/skills/.
  • NEVER auto-activate. Approval is via /dream-review --approve-skill <slug>.
  • Description starts with "Use when…" — this is what the model uses to auto-trigger.
  • Body ≤ 400 LOC. Skills that need more should be decomposed.
  • No destructive ops in body. A skill that says rm -rf is a backdoor; refuse.
  • Per-day cap 3 creations. Prevents instinct/signature noise from flooding the registry.
  • Cite the source. provenance.md records who/what generated the skill (manual / instinct ID / signature hash).

Examples

Example 1 — manual invocation

/create-skill git-cleanup "Use when local branches whose remotes are gone accumulate; prunes them safely."

→ writes .claude/memory.proposed/skills/git-cleanup/{SKILL.md, EXAMPLES.md, triggers.yml, provenance.md}

Example 2 — instinct promotion

/instinct promote inst-014 --as-skill

→ reads .claude/memory/instincts/active.yml entry inst-014; if reinforced ≥ 10, invokes skill-creator with description + examples from the instinct's observations.

Example 3 — auto-detected signature

task-signature-detector finds the same 3-step Bash sequence repeated 6× across 3 sessions → writes a candidate to _candidates.jsonl → next Stop hook invokes skill-creator with that candidate.

Done means

  • .claude/memory.proposed/skills/<slug>/ exists with all 4 files
  • NEXUS YAML handoff emitted
  • Followup task appended to TASKS.md prompting /dream-review --approve-skill <slug>
  • Per-day counter incremented
  • NO file created under .claude/skills/ (that's the approval step)

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.