agentleFS
Sign inStart free
Catherine1401Catherine1401/agent-skills/skills/skill-creator/SKILL.md

skill-creator

Skill2 starsChanged 54 days agoMITView on GitHub

Create or update an agent skill with appropriately scoped instructions and necessary resources. Use for skills; not policies, plugins, or project implementation.

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

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

  • 0

    Common

    Most of the 185 edited versions we read keep these.

  • 5

    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. Distinctive:

    Keep names and descriptions concise and discriminating about capability and applicability.

    line 10
  2. Distinctive:

    Keep skills self-contained.

    line 11
  3. Distinctive:

    Use lowercase letters, digits, and hyphens in names under 64 characters.

    line 14
  4. Very specific:

    Keep purpose, essential constraints, shared workflow, and mode-selection criteria in SKILL.md.

    line 16
  5. Very specific:

    Add scripts/ only to avoid repeatedly rewriting the same logic or when deterministic execution improves reliability.

    line 17
  6. Very specific:

    Use references/ for task-specific information needed in particular contexts.

    line 18
  7. Very specific:

    Use assets/ for files included in generated output.

    line 19
  8. Distinctive:

    Use a bundled validator when available to check frontmatter, naming, and unfinished placeholders.

    line 23
  9. Distinctive:

    When behavioral testing is warranted, verify observable outcomes or meaningful invariants.

    line 24

“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
  • Optimize for agent execution: retain only instructions that change decisions or improve results. Remove prose, headings, examples, alternatives, and files unless they resolve an operational ambiguity. Assume the agent is capable; omit generic advice, repeated rules, and speculative edge cases.
  • Preserve the user's intent, scope, product choices, and authorization boundaries. Do not turn examples, past failures, or preferences into universal requirements, or imply permission to change unrelated configuration or perform additional external actions.
  • For retrying or externally mutating workflows, define a stopping condition proportional to the risk. Approval to complete a task does not expand its scope or execution permissions.
  • Match specificity to risk: describe outcomes and decision criteria for open-ended work; require fixed sequences, parameters, or deterministic scripts only when deviation would cause a concrete correctness, safety, or permissions problem. Preserve non-obvious operational invariants; distinguish requirements from recommendations and local conventions.
  • Keep names and descriptions concise and discriminating about capability and applicability. Add exclusions only to prevent likely misrouting; avoid exhaustive capability lists and catchalls for unrelated tasks.
  • Keep skills self-contained. Depend on another skill or tool only when the workflow genuinely requires it and the target environment provides it. Require specialized review only when requested or genuinely needed.
  • Adapt the work to the request; narrow updates need only focused edits and validation. Ask only when missing information matters and cannot reasonably be inferred; do not request more examples when the task is clear.
  • Respect the user's chosen location and preserve an existing skill's location when updating. For new skills, use a directory discoverable by the target agent; for Codex, default to $CODEX_HOME/skills, or ~/.codex/skills when unset.
  • Use lowercase letters, digits, and hyphens in names under 64 characters; prefer short action-oriented names. Namespace by tool or domain when it improves discovery; name the folder after the skill.
  • Include YAML frontmatter with name and description in SKILL.md. Preserve supported optional fields such as metadata when appropriate; do not change unrelated files or metadata.
  • Keep purpose, essential constraints, shared workflow, and mode-selection criteria in SKILL.md. Move substantial conditional detail into linked resources with reading conditions; do not load every resource by default or duplicate content. Short skills do not need a routing layer.
  • Add scripts/ only to avoid repeatedly rewriting the same logic or when deterministic execution improves reliability. Run new or changed scripts to validate behavior.
  • Use references/ for task-specific information needed in particular contexts; keep one source for each piece of information. For large references, add search terms or a short contents section when useful. Inspect callers and purpose before removing existing resources.
  • Use assets/ for files included in generated output; do not load them as instructions unless inspection is needed. Do not create directories, placeholders, READMEs, installation guides, or auxiliary documentation without a concrete need.
  • For agents/openai.yaml, add UI metadata only when useful and optional interface fields only when the user provides or requests them. If the bundled toolkit includes references/openai_yaml.md, read it before editing; keep values consistent with the skill and preserve unrelated policy, dependencies, and fields.
  • Preserve existing invocation policy unless the user requests a change. New skills allow automatic selection by default; set policy.allow_implicit_invocation: false only when the user explicitly requests explicit-only invocation. Do not infer this mode from sensitive operations or approval requirements; require authorization immediately before the actual mutation.
  • Use a bundled initializer when available and useful for creating a new skill consistently; do not reinitialize existing skills. Request only necessary resources; replace or remove placeholders before finishing. Do not use a generator that replaces all metadata when existing policy or dependencies must be preserved.
  • Use a bundled validator when available to check frontmatter, naming, and unfinished placeholders. Also check description precision, scope, resource discoverability, and script behavior; structural validation does not prove decision quality.
  • When behavioral testing is warranted, verify observable outcomes or meaningful invariants; avoid tests that merely match wording, headings, or regex patterns. Improve from real usage or demonstrated failures; prefer narrow corrections over accumulating universal rules.
  • Use independent subagent evaluation only when complexity or risk makes behavioral validation useful and delegation is available and authorized. Provide a realistic request, the skill, and minimum raw artifacts; withhold intended answers, suspected bugs, and prior conclusions unless needed. Use an isolated temporary workspace and constrain resources and side effects; ask for approval if additional authorization, production impact, or substantial time or cost is involved. Change only what observed outcomes and artifacts support.

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.

Check your setup, free