agentleFS
Cody-SimsCody-Sims/agent-skills/skills/skill-creator/SKILL.md

skill-creator

Skill0 starsChanged 4 months agoMITView on GitHub

Authors, audits, and repairs Agent Skills with clear triggers, concise workflows, progressive disclosure, and valid frontmatter. Use when creating or editing SKILL.md, improving skill efficiency or routing, or fixing skill validation; not for application features or tool-interface design.

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

7 rules. 0 are common, 2 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 185 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

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

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

    When delivery enters scope, use git-and-pr-workflow delivery preflight to inventory the entire authorized base-to-head range before …

    line 33
  2. Distinctive:

    Check catalog descriptions for overlap.

    line 37
  3. Distinctive:

    Keep explicit procedural steps for GPT-based agents.

    line 44
  4. Very specific:

    Keep the core workflow under 500 lines and roughly 5,000 tokens.

    line 46
  5. Very specific:

    Use assets/ for templates.

    line 49
  6. Very specific:

    Use invocation controls only when justified.

    line 56
  7. Very specific:

    Run repository-required validation/tests and gh skill publish --dry-run.

    line 61

“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

Produce or repair skills/<name>/SKILL.md with reliable routing, a bounded workflow, and validated portable frontmatter. Improve existing skills before adding overlapping ones.

Inputs

Require the capability and a realistic activating request, or an existing skill and its observed problem. Ask only for missing consequential scope.

Workflow

  1. Identify the mode: read-only audit, create, routing repair, validation repair, or efficiency improvement. An audit returns cited findings and proposed changes, not edits, version bumps, installation, publication, or authentication setup. Steps 7-8 apply to authorized edits only. For an existing skill, read it and relevant neighbors; preserve its name and unrelated resources. Do not restart the full creation checklist. For session-informed optimization, read session audit before querying history. Patch an observed gap, not a rule already repaired in the current or installed release. When delivery enters scope, use git-and-pr-workflow delivery preflight to inventory the entire authorized base-to-head range before readiness/install work, not only the current edits.
  2. Define capability, activating requests, inputs, stop condition, and output. Check catalog descriptions for overlap; extend an existing capability or make neighboring boundaries explicit. Read full neighbors only when needed.
  3. For a new or changed description, add at least three positive prompts and two near-misses. Tighten false-positive triggers; add missing user vocabulary for false negatives without broadening the capability.
  4. Front-load concrete trigger words in a concise third-person description with an explicit "Use when" clause. Put routing signals here, not only in the body. Keep explicit procedural steps for GPT-based agents; remove repeated rationale and ceremony, not safety rules, acceptance criteria, or escalation conditions.
  5. Keep the core workflow under 500 lines and roughly 5,000 tokens. Include only needed actions and clear stop/output contracts. Link conditional detail in references/ directly from the skill; read it only for its stated condition. Use assets/ for templates. Prefer prose for judgment; add a script only for deterministic, repetitive, fragile, or machine-verifiable work, stating whether to run or read it.
  6. Validate frontmatter against the repository contract: name equals directory, lowercase letters/digits/single hyphens, at most 64 characters, no reserved names; description meets repository length limits and has no angle brackets. Include license: MIT and quoted metadata version, author, and tier. Use invocation controls only when justified; never add allowed-tools, when_to_use, tools, model, or paths. Preserve supported runtimes.
  7. Bump semantic versions for changes and update directly dependent registry, pack, and documentation metadata. Credit adapted concepts without copying wording; verify reuse rights. Do not commit private session evidence.
  8. Run repository-required validation/tests and gh skill publish --dry-run. Inspect outputs and exit status; resolve every relative link and referenced file. Fix reported issues and rerun affected checks, not unchanged gates simply because another handoff occurred.
  9. For efficiency claims, compare the previous and candidate skill on matched tasks with the same target runtime/model and correctness rubric. Record calls, output/context size, elapsed time, and available usage separately. Static size reductions or fixture smoke tests do not prove model uplift or billed-token savings. Honor contribution-review gates before publishing.

Resources

  • Frontmatter reference: consult for field rules.
  • Description patterns: consult for routing examples or collisions.
  • Skill template: use for a new skill, not a repair.
  • Session audit: use for historical failures, repetitive work, or research-backed workflow optimization.
  • Event-audit helper: run, rather than read, only for authorized raw-log analysis when indexed evidence is insufficient.

Output

For an audit, report coverage, cited findings, proposed changes, and evidence limits without implying mutation. For edits, report changed paths, final descriptions, line counts, validation results, and unresolved assumptions. Include measured improvements and their limits; never leave the handoff empty. Generate runtime copies through the supported installer, not hand-maintained edits.

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.