agentsclimarketplace
Sign in

Skill curator

Skill justinramos101/agent-skill-kit/.agents/skills/skill-curator

Use when turning source material (books, talks, articles, notes) or observed agent behavior into a new published skill for this catalog. Do not use for reviewing an already-built skill for release readiness; use `skill-reviewer` for that.From its SKILL.md

Install
npx -y skills add justinramos101/agent-skill-kit --skill skill-curator

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.

SKILL.md

2.4 KB, 548 tokens by cl100k_base, as published. Nobody here has run it

Skill Curator

Turn source material into a published skill that earns its context cost. Run autonomously: research, deduplicate, scaffold, validate, hand off — stating assumptions in the output. Stop to ask only for rights/safety ambiguity or a genuine design tie.

House style (invariants)

  • Flat by default. One SKILL.md under ~500 words plus a small set of references, one per branch that exceeds a screen. Add routing CSVs only when two axes genuinely multiply — the published catalog currently has none.
  • AFK contract. No mode menus, no bare-invocation menus, no wait-for-user gates. The flat static-check shape rejects them.
  • Gotchas, not manuals. Hold every line to docs/skill-authoring-principles.md: routing-first description, inlined invariants, named heuristics, deletion test, stop conditions for loops.
  • Paraphrase boundary. Never reproduce distinctive source text; provenance lives in skill.json.inspired_by, never in SKILL.md.

Workflow

  1. Research. Build a dossier per references/research-dossier.md; save it to .agents/state/source-dossiers/<slug>.md. Run its paraphrase audit.
  2. Deduplicate. Grep the catalog for overlap. If the idea belongs inside an existing skill, extend that skill instead of minting one.
  3. Scaffold skills/<name>/ per AGENTS.md §Per-skill required artifacts: SKILL.md; skill.json (status: "published", resolvable @maintainers, description exactly equal to the frontmatter description, pack tags in tags — never a source title, and metadata.{family,function,catalog_summary,capabilities}); the generated README.md (never hand-edited — run python3 scripts/build-catalog.py --write); and evals/ with run-static-checks.sh (flat shape via scripts/check-skill-static.py), trigger-evals.json, activation-cases.md.
  4. Validate. Run just check and self-review against the authoring principles. Fix and rerun; cap at 3 rounds, then report what is still red and why.
  5. Hand off to skill-reviewer with the dossier path, known risks, and stated assumptions.

What ships with it: 3 files

5.0 KB alongside SKILL.md, 1 of them executable

evals/

references/

Keep looking

Skills are one crate of 326,367. 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.