agentleFS
sunitghubsunitghub/canon-skills/skills/skill-export/SKILL.md

skill-export

Skill9 starsChanged 21 days agoMITView on GitHub

Exports any flat canon skill as a paste-ready prompt, or refines an existing prompt using efficiency.md standards. Use when running a skill in claude.ai, a system prompt, or another environment that doesn't load SKILL.md natively. Invoke as skill-export <skill-name> or skill-export <skill-name> "<prompt>". Rejects skills with sub-skills.

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 6 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Very specific:

    Refine mode: If a second argument is present — a quoted string ("…" or '…') after the skill name — enter refine mode:

    line 15
  2. Very specific:

    Remove hedging and filler — the canonical no-op/filler list lives in tools/no-op-phrases.txt (shared with canon-dev.sh lint and …

    line 19
  3. Distinctive:

    Remove preamble paragraphs that restate the task before giving instructions — start with the first instruction

    line 20
  4. Distinctive:

    Do not merge sentences that express different constraints — each distinct instruction must survive in the output

    line 23
  5. Distinctive:

    Before writing output, enumerate every distinct instruction in the raw prompt.

    line 24
  6. Distinctive:

    Stop — do not continue to Step 1.

    line 30
  7. Distinctive:

    If any directory is returned, the skill has sub-skills and cannot be reliably distilled.

    line 42
  8. Very specific:

    Strip frontmatter: Read the SKILL.md. Remove everything between and including the opening --- and closing --- frontmatter delimiters (the …

    line 48
  9. Distinctive:

    If the skill body begins with blank lines after the frontmatter, preserve them.

    line 68

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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 Export

Reads a canon skill and outputs its content as clean, paste-ready text — no frontmatter, no SKILL.md structure. Use when you need to run a skill in claude.ai, a system prompt, or any environment that doesn't load SKILL.md files natively.

Steps

  1. Refine mode. If a second argument is present — a quoted string ("…" or '…') after the skill name — enter refine mode:

    1. Take the quoted string as the raw prompt text.
    2. Apply standards/efficiency.md principles to produce a tighter version:
      • Remove hedging and filler — the canonical no-op/filler list lives in tools/no-op-phrases.txt (shared with canon-dev.sh lint and repo-workflow-audit); it includes "please", "make sure to", "you should", "always remember to", "it's important to", "note that", "keep in mind that"
      • Remove preamble paragraphs that restate the task before giving instructions — start with the first instruction
      • Collapse multiple sentences that express the same constraint into one direct instruction
      • Imperative mood, active voice throughout
      • Do not merge sentences that express different constraints — each distinct instruction must survive in the output
    3. Before writing output, enumerate every distinct instruction in the raw prompt. Verify each has a counterpart in the refined version. If any would be dropped, retain it.
    4. Output the refined prompt with no preamble or wrapper. Then append:
      ---
      Source: refined from user input (standards/efficiency.md)
    5. Stop — do not continue to Step 1.
  2. Locate the skill. Find skills/<name>/SKILL.md relative to the repo root. If the file does not exist, output:
    skill-export: '<name>' not found — run ./tools/skills.sh list to see available skills.
    Then stop.
  3. Flat check. Run:
    find skills/<name> -maxdepth 1 -mindepth 1 -type d -not -name "evals"
    If any directory is returned, the skill has sub-skills and cannot be reliably distilled. Output:
    skill-export: '<name>' has sub-skills — not supported. Use Claude Code with skills.sh for full fidelity.
    Then stop.
  4. Strip frontmatter. Read the SKILL.md. Remove everything between and including the opening --- and closing --- frontmatter delimiters (the name:, description:, category:, tags:, hidden: block at the top). Do not remove any --- horizontal rules that appear later in the body.
  5. Output. Print the following, with no added commentary:

    # <name> — paste into Project instructions or system prompt

    Then the stripped skill content verbatim — all headings, lists, code blocks, and prose exactly as written.

    Then a footer:

    ---
    Source: skills/<name>/SKILL.md

Rules

  • Output only the skill content. No preamble, no explanation, no "here is the exported skill" wrapper.
  • Preserve every heading level, list item, and code block exactly. Do not reword, summarize, or reformat.
  • If the skill body begins with blank lines after the frontmatter, preserve them.

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.