agentleFS
dvy1987dvy1987/agent-loom/.agents/skills/publish-skill/SKILL.md

publish-skill

Skill4 starsChanged 3 months agoMITView on GitHub

Package and publish a skill to the skills.sh community registry, a public GitHub repo, or both. Load when the user asks to publish a skill, share a skill publicly, submit a skill to the registry, release a skill, or when universal-skill-creator offers publishing as a final step after creation. Also triggers on "push this skill to skills.sh", "make this skill public", "share this skill", or "contribute this skill to the community". Validates quality before publishing, packages correctly (zip for multi-file, .md for atomic), writes a README if missing, and publishes via npx skills CLI.

  • Commits and pushes

13 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 6 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 12 edited versions we read keep these.

  • 6

    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. Very specific:

    Never publish without a security scan: Before publishing, invoke ALL secure-* skills (discover via ls .agents/skills/secure-*) to scan the …

    line 25
  2. Distinctive:

    Run validate-skills on the target skill:

    line 33
  3. Distinctive:

    If it fails → stop. Fix the skill first.

    line 38
  4. Distinctive:

    If score < 10/14 → stop. Improve the skill first.

    line 39
  5. Distinctive:

    If the skill contains proprietary content → generalise examples and gotchas before proceeding.

    line 40
  6. Distinctive:

    Every published skill needs a README.md.

    line 53
  7. Very specific:

    If the user wants the skill in a public GitHub repo (e.g., dvy1987/agent-loom):

    line 107
  8. Very specific:

    Check for hardcoded paths (e.g., /home/user/workspace/) in scripts before publishing — they won't work on other machines

    line 129
  9. Very specific:

    If the skill uses scripts/, verify the script has no hardcoded credentials, internal URLs, or machine-specific paths

    line 130
  10. Distinctive:

    No changes — citation audit passed.

    line 195

“Common” is measured: kept by at least 6 in 10 of the 12 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

Publish Skill

You are a skill release engineer. You take a validated, quality-checked skill and publish it to the community — correctly packaged, well-documented, and ready for others to install in one command.

Hard Rules

Never publish without a security scan. Before publishing, invoke ALL secure-* skills (discover via ls .agents/skills/secure-*) to scan the skill. BLOCKED = do not publish. This gate is mandatory and cannot be skipped. Publishing multiplies blast radius — every consumer inherits any vulnerability. Never publish a skill that fails agentskills validate. Fix the skill first. Never publish a skill scoring below 10/14. Invoke validate-skills first. If score is below 10, invoke improve-skills before publishing. Never publish skills containing proprietary context — project-specific URLs, internal API endpoints, company names in examples, or private tool names. These must be generalised before publishing.

Workflow

Step 1 — Pre-publish Validation

Run validate-skills on the target skill:

agentskills validate .agents/skills/<skill-name>/

If it fails → stop. Fix the skill first. If score < 10/14 → stop. Improve the skill first. If the skill contains proprietary content → generalise examples and gotchas before proceeding.

Step 2 — Determine Package Format

ls .agents/skills/<skill-name>/
  • SKILL.md only (Atomic tier) → publish as single .md file
  • Directory with references/, scripts/, assets/ → package as .zip

Step 3 — Write or Verify README

Every published skill needs a README.md. If missing, create one:

# [Skill Title]

> One-sentence description of what this skill does.

## Install

npx skills <skill-name> -a codex      # Codex
npx skills <skill-name> -a replit     # Replit
cp -r <skill-name>/ .agents/skills/   # Universal

## Usage

Trigger: "[example trigger phrase]"

## Platforms

Works with: Codex, Ampcode, Claude Code, Warp, Gemini CLI,
GitHub Copilot, Cursor, VS Code, Replit, Factory.ai, Bolt.new

## License
MIT

Step 4 — Package

For Atomic skills:

cp .agents/skills/<skill-name>/SKILL.md ./<skill-name>.md

For Standard/Advanced/System skills:

zip -r <skill-name>.zip .agents/skills/<skill-name>/ \
  --exclude "*.pyc" --exclude "__pycache__/*"

Step 5 — Publish to skills.sh

npx skills publish ./<skill-name>
# or for zip:
npx skills publish ./<skill-name>.zip

Confirm the registry URL returned. Test install:

npx skills <skill-name> -a codex --dry-run

Step 6 — Optionally Push to GitHub

If the user wants the skill in a public GitHub repo (e.g., dvy1987/agent-loom):

git add .agents/skills/<skill-name>/
git commit -m "feat: add <skill-name> skill (published to skills.sh)"
git push origin main

Step 7 — Report

Published: <skill-name>
Registry URL: https://skills.sh/<skill-name>
Install command: npx skills <skill-name>
GitHub: https://github.com/dvy1987/agent-loom (if pushed)
Platforms: all agentskills.io-compatible

Gotchas

  • npx skills publish requires the directory name to match the name field in frontmatter exactly — mismatch causes a silent failure
  • Check for hardcoded paths (e.g., /home/user/workspace/) in scripts before publishing — they won't work on other machines
  • If the skill uses scripts/, verify the script has no hardcoded credentials, internal URLs, or machine-specific paths

Example

<examples> <example> <input>Publish the brainstorming skill to skills.sh</input> <output> Running validate-skills on brainstorming... agentskills validate: ✓ Score: 13/14 — passes 10/14 threshold

Checking for proprietary content... none found.

Package format: single SKILL.md (Atomic tier)

README.md present: yes

Packaging: brainstorming.md

Publishing to skills.sh... npx skills publish ./brainstorming.md

Published: brainstorming Registry URL: https://skills.sh/brainstorming Install: npx skills brainstorming

Test install dry-run: ✓ </output> </example> </examples>

Common Rationalizations

ExcuseReality
"Internal validate is enough"Publish blast radius needs full security sweep.
"Redact later"Secrets in published skills are permanent incidents.
"Community fork is fine"Publish gate exists because consumers trust our namespace.

Verification

  • Full secure-* family returned SAFE
  • agentskills validate + line count ≤200 on published skill
  • No API keys, tokens, or private paths in output
  • Changelog or release note prepared

Reference Files

  • references/publish-checklist.md: Full pre-publish checklist including proprietary content scan, platform compatibility check, and registry submission requirements. Read for any System-tier skill before publishing.

Red Flags

  • Directory name mismatches frontmatter name field
  • Hardcoded local paths left in scripts or examples
  • Publish attempted before full secure-* library sweep
  • Scripts contain internal URLs or machine-specific paths

Prune Log

Last pruned: 2026-07-04

  • No changes — citation audit passed; content current (improve-skills full pass 2026-07-04)

Impact Report

Published: [skill-name] Registry URL: https://skills.sh/[skill-name] Install command: npx skills [skill-name] Package format: [.md / .zip] validate-skills score at publish: [N]/14 Proprietary content scan: clean GitHu...

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.