agentsclimarketplace

Custom skill creator

Skill pngdeity/apm-user-repository/.agents/skills/custom-skill-creator

Personal APM marketplace — skills, prompts, agents, and instructions for AI coding agents

Install
npx -y skills add pngdeity/apm-user-repository --skill custom-skill-creator

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Meta-skill for creating and packaging new AI agent skills. Use when requested to create a skill, document a multi-step process, or configure agent behavior as a procedural skill module. Not for single atomic functions or global architecture constraints — those go in AGENTS.md.

SKILL.md

4.6 KB, as published. Nobody here has run it

Skill: Custom Skill Creator (custom-skill-creator)

This is the authoritative Meta-Skill for the pngdeity workspace. It extends the built-in skill creation process by enforcing the strict architectural boundaries of the "Abstraction Stack" and prioritizing validation and optimization based on agentskills.io standards.

Usage

Run this skill whenever requested to "create a skill," "document a process," or "configure agent behavior."

Phase 1: The Decision Heuristic (The Gate)

Before writing any files, you MUST evaluate the user's request against this logic. Do not create a SKILL.md if the request fails this gate.

  1. Global Constraint/Architecture? -> Write it in AGENTS.md. (Refuse skill creation).
  2. Single, Atomic Function? -> This is a Tool. Suggest a code implementation. (Refuse skill creation).
  3. Specialized, Multi-Step Workflow? -> This is a Skill. Proceed to Phase 2.

Phase 2: Skill Initialization & Drafting

  1. Create the skill directory manually:
    • mkdir -p skills/<skill-name>/references skills/<skill-name>/scripts
    • Create SKILL.md with the YAML frontmatter template below.
    • If directory creation fails: Check write permissions and parent directory existence. Fall back to a single SKILL.md file without subdirectories if needed.
  2. Read references/agentskills-standards.md for frontmatter field rules and references/spec-authority.md for the canonical specification sources (agentskills.io, Microsoft Agent Framework, Anthropic Claude).
  3. Draft the SKILL.md and any necessary scripts/references. Keep the main body under 500 lines (Progressive Disclosure).
  4. Validate structure: skills-ref validate ./skills/<skill-name> (or npx skills-ref validate from the agentskills.io reference library).

Phase 3: Mandatory Validation

A skill is not complete until its behavior is validated.

  1. Define explicit test cases or dry-run instructions within the SKILL.md (e.g., "Verification Step: Run script X and ensure output matches Y").
  2. If the skill relies on executable scripts/, you MUST execute them locally to ensure they return LLM-friendly stdout (no massive tracebacks).

Phase 4: Optimization & Security (Post-Stability)

Once the SKILL.md is structurally sound and validated:

  1. Trigger Optimization: Review the YAML description. It must function as an API doc for the agent router. Rewrite it to define exact trigger conditions.
  2. Constraint Hardening: Ensure instructions use imperative mood and handle edge cases without hallucinating.
  3. Regenerate skill-index.json: Run node skills/verification/generate-skill-index.cjs to rebuild the catalog index with updated tags and compatibility tokens from all SKILL.md files.
  4. Security Review: Per the agentskills.io specification and Microsoft Agent Framework guidance (https://learn.microsoft.com/en-us/agent-framework/agents/skills):
    • Review content: Read all skill files — instructions must not attempt to bypass safety guidelines, exfiltrate data, or modify agent configuration.
    • Verify scripts: If scripts/ contains executables, confirm their behavior matches stated intent. Run them in an isolated environment first.
    • Check provenance: Only package skills from trusted sources. Prefer skills with version control history and active maintenance.
    • Audit trail: Ensure the skill's name, description, and compatibility fields are accurate and would not mislead a router.
    • If any security concern is found: Do not package. Document the finding and recommend remediation.

Packaging

Once Phase 4 is complete, make the skill discoverable:

  1. Validate: skills-ref validate ./skills/<skill-name> (or npx skills-ref validate if not installed globally).
    • If skills-ref is unavailable: Skip validation and note the gap. Proceed with structural verification (step 4) instead.
  2. Register for cross-client discovery: ln -sf <absolute-path-to-skill> ~/.agents/skills/<skill-name>
  3. Run node skills/verification/skill-compliance-check.cjs to confirm all checks pass.

Gives 1 of the 12 instructions most skill authoring skills give

Counted across 521 of the 523 authors here whose files we hold, read 2026-08-06

  • keep skill files under 500 linesin 182 of 521, across 89 files
  • use imperative form in instructionshere, and in 101 of 521, across 30 files
  • draft assertions while test runs are in progressin 88 of 521, across 22 files
  • save test cases to evals jsonin 87 of 521, across 21 files
  • create two to three realistic test promptsin 85 of 521, across 20 files
  • write skill descriptions to be pushyin 84 of 521, across 19 files
  • ask questions about edge cases and input formatsin 81 of 521, across 16 files
  • save timing data immediately when runs completein 74 of 521, across 9 files
  • include all trigger conditions in the skill descriptionin 73 of 521, across 7 files
  • capture intent before writing a skillin 70 of 521, across 4 files
  • launch all test runs in a single turnin 68 of 521, across 2 files
  • write the description in third personin 56 of 521, across 19 files

Said here and by no other author read

  • refuse skill creation for global constraints
  • refuse skill creation for single atomic functions
  • write global constraints to AGENTS.md
  • suggest code implementation for atomic functions
  • define explicit test cases in SKILL.md
  • execute local scripts to verify output

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.