agentleFS
nexuschainsnexuschains/operator-os/.claude/skills/skill-creator/SKILL.md

skill-creator

Skill0 starsChanged 5 months agoMITView on GitHub

Scaffolds a new conformant skill under .claude/skills/, including the folder, SKILL.md frontmatter (name, description, model pin), body structure, and optional references/ directory. Invoke after find-skills confirms no existing skill covers the need, or when an operator asks to "create a skill for X", "add a skill that does Y", or "build a skill to automate Z". Also handles registering the new skill's model pin in docs/conventions/skill-model-pins.md as a mandatory part of creation.

We split this skill into the rules it gives an agent

14 rules. 0 are common, 7 are distinctive, and 7 are very specific to this project.

  • 0

    Common

    Most of the 183 edited versions we read keep these.

  • 7

    Distinctive

    This author's own choices. Worth a look.

  • 7

    Very specific

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

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

    Prerequisite: run find-skills first. If a matching skill exists, use it or compose it — do not author a duplicate.

    line 11
  2. Distinctive:

    Before writing anything, gather:

    line 17
  3. Distinctive:

    If the operator has not provided all four, ask for what is missing before proceeding.

    line 27
  4. Very specific:

    No reserved words: anthropic, claude

    line 38
  5. Very specific:

    No vague names: helper, utils, tools, documents, data, files

    line 39
  6. Distinctive:

    Must match the folder name exactly

    line 40
  7. Distinctive:

    Write the description to be slightly specific and trigger-oriented — it is the primary mechanism by which the harness decides to invoke …

    line 103
  8. Very specific:

    If the skill body would exceed 500 lines, or if there are three or more supporting documents, create a references/ subdirectory:

    line 109
  9. Distinctive:

    Verify before handing off to the operator:

    line 139
  10. Distinctive:

    Do not scaffold a skill for a one-off task.

    line 161
  11. Distinctive:

    Do not skip the find-skills prerequisite check.

    line 162
  12. Very specific:

    Do not leave placeholder text in the generated SKILL.md.

    line 163
  13. Very specific:

    Do not use inherit as a model pin without stating a clear reason why the caller's model is appropriate here.

    line 164
  14. Very specific:

    Do not create nested reference chains (SKILL.md → a.md → b.md → content).

    line 165

“Common” is measured: kept by at least 6 in 10 of the 183 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

Scaffolds a new operator-os-conformant skill from a plain-language description of the need. The output is a populated SKILL.md ready for operator review, with the model pin registered.

Prerequisite: run find-skills first. If a matching skill exists, use it or compose it — do not author a duplicate.


Step 1 — Capture intent

Before writing anything, gather:

  1. What the skill does — the action it takes (verb-first: "scans", "generates", "validates")
  2. When to invoke it — the trigger phrases, situations, or conditions that call for this skill
  3. What it produces — the artifact or state change that results
  4. Model tier — apply the decision tree from docs/conventions/model-orchestration.md:
    • Haiku: purely mechanical work (template fill, regex/JSON patching, config scaffolding, counting)
    • Sonnet: judgment-bearing work (merge decisions, style review, classification, scoring)
    • Opus: architectural/strategic work (novel synthesis, multi-source reasoning, designing new schemas)

If the operator has not provided all four, ask for what is missing before proceeding. Do not guess at model tier — apply the decision tree explicitly and state the rationale.


Step 2 — Choose a name

Apply the naming rules from docs/conventions/skill-conventions.md:

  • Gerund form preferred: processing-pdfs, analyzing-sources
  • Noun-phrase form acceptable: pdf-processor
  • Lowercase, hyphens only, max 64 characters
  • No reserved words: anthropic, claude
  • No vague names: helper, utils, tools, documents, data, files
  • Must match the folder name exactly

Propose the name and explain the choice. Get confirmation before creating the folder.


Step 3 — Scaffold the folder and SKILL.md

Create:

.claude/skills/<skill-name>/
└── SKILL.md

SKILL.md template:

---
name: <skill-name>
description: <third-person, ≤1024 chars, states WHAT it does + WHEN to invoke it>
model: <haiku | sonnet | opus | inherit>
---

# <Skill Title>

<One-sentence summary of what this skill does and why it exists.>

---

## When to invoke

<Bullet list of trigger conditions, phrases, and situations. This is the second place the model looks when routing — be specific.>

---

## What to do

<Numbered steps. Imperative form. Each step has a clear action and a clear output. No vague "consider" language.>

---

## What not to do

<Hard stops. Things that look like they belong here but do not. Common wrong turns.>

---

## Related

<Links to related skills, protocol docs, or convention files. Forward slashes in all paths.>

Fill every section with content derived from Step 1. Do not leave placeholder text.

Description field rules

The description must:

  • Be third person only ("Scans and reports..." not "I can help you..." or "You can use this...")
  • State both WHAT the skill does AND WHEN to invoke it (both conditions in one field)
  • Be ≤ 1024 characters
  • Contain no XML tags

Write the description to be slightly specific and trigger-oriented — it is the primary mechanism by which the harness decides to invoke this skill.


Step 4 — Add references/ if needed

If the skill body would exceed 500 lines, or if there are three or more supporting documents, create a references/ subdirectory:

.claude/skills/<skill-name>/
├── SKILL.md
└── references/
    ├── workflow.md
    └── schema.md

Link from SKILL.md using forward-slash paths: references/workflow.md. Keep the chain one level deep — no nested references.

Reference files longer than 100 lines must include a TOC at the top.


Step 5 — Register the model pin (mandatory)

Open docs/conventions/skill-model-pins.md and append a row to the Catalog table:

| `<skill-name>` | <haiku\|sonnet\|opus\|inherit> | <one-sentence rationale citing the decision tree> |

This step is not optional. A skill without a pin row is a lint violation. Do this in the same operation as creating the SKILL.md.


Step 6 — Run the authoring checklist

Verify before handing off to the operator:

- [ ] name: lowercase, hyphens, ≤64 chars, no reserved words, matches folder name
- [ ] description: third person, ≤1024 chars, states WHAT + WHEN, no XML tags
- [ ] model: pin declared
- [ ] Row added to docs/conventions/skill-model-pins.md
- [ ] SKILL.md body ≤ 500 lines
- [ ] Three or more bundled docs are inside references/
- [ ] All reference files one level deep from SKILL.md
- [ ] Forward slashes in all paths
- [ ] MCP tools fully-qualified if present (ServerName:tool_name)
- [ ] No time-sensitive content in main body
- [ ] No vague names, no first-person copy, no XML in name or description

Report any violations and fix them before handing off.


What not to do

  • Do not scaffold a skill for a one-off task. Skills justify maintenance cost only through repeated use. If the operator is doing something once, do it inline.
  • Do not skip the find-skills prerequisite check.
  • Do not leave placeholder text in the generated SKILL.md. Every section must be populated.
  • Do not use inherit as a model pin without stating a clear reason why the caller's model is appropriate here.
  • Do not create nested reference chains (SKILL.md → a.md → b.md → content). Keep references one level deep.

Related

  • find-skills — always run before authoring to confirm no duplicate exists
  • docs/conventions/skill-conventions.md — full authoring rules; this skill operationalizes them
  • docs/conventions/model-orchestration.md — decision tree for choosing a model pin
  • docs/conventions/skill-model-pins.md — the pin catalog this skill must update
  • docs/protocols/capability-extension.md — the self-improvement flywheel context

Download now (ZIP)

  • Other versions188 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 188

    1. moltbot/moltbot★ 393ksame name
    2. clawdbot/clawdbot★ 393ksame name
    3. steipete/warelay★ 392ksame name
    4. moltbot/clawdbot★ 392ksame name
    5. steipete/clawdis★ 392ksame name
    6. molt-bot/clawdbot★ 391ksame name
    7. openclaw/openclaw★ 391ksame name
    8. anthropics/skills★ 186ksame name
    9. openai/codex★ 133ksame name
    10. google-gemini/gemini-cli★ 109ksame name

    178 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.

  • Used alongsideProjects with it also use taskflow-inbox-triage, community-led-growth 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 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

    • nicepkgSkill

      ph-community-outreach

      Engage Reddit, Indie Hackers, Hacker News and other communities for Product Hunt launches. Use this skill to identify relevant communities, craft appropriate posts, and drive cross-platform engagement without appearing spammy.

      Also in 5 projects that use this

    • seedprodSkill

      session-logs

      Search and analyze your own session logs (older/parent conversations) using jq.

      Also in 18 projects that use this

    • AnthropicSkill

      academy-guide

      Stop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content.

      Also in 5 projects that use this

    • ArogyaReddySkill

      skill-share

      A skill that creates new Claude skills and automatically shares them on Slack using Rube for seamless team collaboration and skill discovery.

      Also in 23 projects that use this

    • +3 more from nicepkg
  • In this repository11 other files it gives its agents.

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

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.