本文へ移動
cccskills
無料GitHub で公開

publish-skill

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.

インストール方法を見る

含まれるファイル(3)

  • SKILL.md6.2 KB
  • references/examples.md1.8 KB
  • references/publish-checklist.md2.0 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

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 BLOCKED = do not publish.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...

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Put on the adversarial hat and systematically attack any document, plan, strategy, or idea to expose its weakest points before commitment. Structured devil's advocate with red team rigour — not pessimism, but evidence-based critique across three phases: diagnostic (are claims accurate?), creative (is the problem artificially constrained?), challenge (are solutions robust?). Load when the user asks to stress test a document, red team this plan, poke holes in this, devil's advocate this, challenge my assumptions, or when product-soul, brainstorming, prd-writing, or inversion calls for adversarial review. Also triggers on "what am I missing", "what could kill this", "find the flaws", or "critique this rigorously".

日本語の概要は準備中です。原文の説明を表示しています。

dvy1987/agent-loom32026年8月8日 更新

Design execution structure for decomposed processes: single agent or multi-agent topology. Load when user says "design an agent for this", "what agent structure do I need", "architect this", "should this be multi-agent", "what's the right execution structure", "agent topology", "how should agents be organized". Takes process-decomposer output as primary input. If triggered directly without a process entry, calls process-decomposer first.

日本語の概要は準備中です。原文の説明を表示しています。

dvy1987/agent-loom32026年8月8日 更新

Internal skill. Called by setup-evaluation after a PASS. Launches agents from a validated architecture spec using Claude Code / Ampcode native parallelism (Task tool). Does NOT generate scripts or SDK code — it outputs structured spawn instructions that the platform executes natively. Never invoked directly by the user. Never launches without a setup-evaluation PASS.

日本語の概要は準備中です。原文の説明を表示しています。

dvy1987/agent-loom32026年8月8日 更新

Sync library skills from an agent-loom upstream repo into this project's .agents/skills while preserving project-local and forked skills. Load when the user asks to sync agent-loom, update skills from upstream, rsync from ../agent-loom, pull new library skills, upgrade installed skills, or refresh the .agents folder without losing custom project skills. Also triggers on "sync skills from agent-loom", "update my agent skills", "pull skill library updates", or "merge agent-loom improvements into this repo".

日本語の概要は準備中です。原文の説明を表示しています。

dvy1987/agent-loom32026年8月8日 更新

Instrument a shipped product's AI agents with tracing and observability so you can see what they did, why outputs happened, and what each run cost. Plain-language primer plus free-tier-first backend selection (Langfuse, Phoenix, LangSmith, Braintrust) and OpenTelemetry/OpenInference instrumentation. Load when the user asks to add observability, add tracing, instrument my agents, see what my agent is doing in production, set up Langfuse or Phoenix or LangSmith, debug why my agent gave a bad answer, or track LLM cost per request. Also fires when agent-system-architecture or setup-evaluation requires an observability plan for an agent-chain product. NOT for tracing the coding agent itself — that is run-trace. Precondition for runtime-learning-loop.

日本語の概要は準備中です。原文の説明を表示しています。

dvy1987/agent-loom32026年8月8日 更新

Run a structured retrospective after development-phase runs of your product's agents — interview the owner in plain language about what went well and poorly, draft ranked improvement hypotheses, then design and run small n=1/n=2 experiments with pre-declared success criteria, guardrails, stop conditions, and a cost/ROI kill-switch. Load when the user says how did that run go, retro this run, the agent output was bad, what should we improve, draft hypotheses, run a small experiment, or after repeated dev runs of an agentic system produce uneven quality. Priority: output quality over performance over cost, each with diminishing-returns stops. NOT a product A/B test (experimentation), NOT coding-agent harness repair (harness-evolution), NOT production-scale learning (runtime-learning-loop).

日本語の概要は準備中です。原文の説明を表示しています。

dvy1987/agent-loom32026年8月8日 更新

dvy1987 のスキルをすべて見る

このスキルの問題を報告する