How do I install the Cross Agent Skill Publishing skill?

Run npx skillmds@latest add publieople/cross-agent-skill-publishing in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

What does the Cross Agent Skill Publishing skill do?

Publish SKILL.md for any agent: skills.sh, CLI-first. It is listed under AI & ML on SkillMD.

Is Cross Agent Skill Publishing safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Cross Agent Skill Publishing?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Cross Agent Skill Publishing free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Cross Agent Skill Publishing?

publieople (@publieople) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Cross Agent Skill Publishing by publieople

Name: Cross Agent Skill Publishing
Author: publieople

File contents

Cross-Agent Skill Publishing

Making a project's capability consumable by ANY AI agent (Claude Code, Cursor, Codex, Gemini, Hermes, …). Session 2026-08-02, BootKeeper — the user's chosen standard is skills.sh (Vercel).

The core problem

Skill formats are NOT cross-agent: Claude SKILL.md, Hermes skill, OpenCode differ in packaging. But every agent can run a CLI. Therefore:

A CLI is the truly agent-agnostic base; SKILL.md is just the instruction sheet over it.

MCP can be added later as another thin adapter over the SAME CLI — never bake the AI interface into the core. This is the pattern the user endorsed for BootKeeper: stable bootkeeper CLI + root SKILL.md; MCP deferred as an adapter.

skills.sh (Vercel) — the standard

vercel-labs/skills (~28k stars). Format = standard SKILL.md, no special header.
Install on any supporting agent: npx skills add <owner>/<repo>
Supports 19+ agents incl. nous-research (Hermes), Claude Code, Cursor, Codex, GitHub Copilot, Windsurf, Gemini, Cline.
Badge for README: [![skills.sh](https://skills.sh/b/owner/repo)](https://skills.sh/owner/repo)
Hermes page: https://skills.sh/agent/nous-research

Placement & structure that works

SKILL.md at repo root — that's what npx skills add picks up.
Self-contained: the agent has zero other context. Include:
- When to use (trigger)
- Every command with examples
- Output format (JSON field names)
- Platform notes (e.g. "must run on Windows, else returns []")
- Agent rules (never fabricate; risk is authoritative; writes not yet available)

Worked example (BootKeeper)

publieople/BootKeeper: root SKILL.md documents bootkeeper list/get/analyze/ snapshot (JSON out). CLI builds cross-platform; SKILL.md documents JSON shapes an agent parses directly.

Pitfalls

Don't skip the CLI. A doc-only SKILL.md gives the agent nothing to run — it would write its own fragile reimplementation each time.
Don't bake MCP into core. MCP is an adapter; the CLI is the contract.
Verify install format: if npx skills add fails frontmatter validation, fall back to git clone (see hermes-external-skill-install skill).

---name: cross-agent-skill-publishingdescription: Publish SKILL.md for any agent: skills.sh, CLI-first.---# Cross-Agent Skill PublishingMaking a project's capability consumable by ANY AI agent (Claude Code, Cursor,Codex, Gemini, Hermes, …). Session 2026-08-02, BootKeeper — the user's chosenstandard is skills.sh (Vercel).## The core problemSkill formats are NOT cross-agent: Claude SKILL.md, Hermes skill, OpenCode differin packaging. But **every agent can run a CLI**. Therefore:> **A CLI is the truly agent-agnostic base; SKILL.md is just the instruction> sheet over it.**MCP can be added later as another thin adapter over the SAME CLI — never bakethe AI interface into the core. This is the pattern the user endorsed forBootKeeper: stable `bootkeeper` CLI + root SKILL.md; MCP deferred as an adapter.## skills.sh (Vercel) — the standard- `vercel-labs/skills` (~28k stars). Format = standard SKILL.md, no special header.- Install on any supporting agent: `npx skills add <owner>/<repo>`- Supports 19+ agents incl. **nous-research (Hermes)**, Claude Code, Cursor, Codex, GitHub Copilot, Windsurf, Gemini, Cline.- Badge for README: `[![skills.sh](https://skills.sh/b/owner/repo)](https://skills.sh/owner/repo)`- Hermes page: `https://skills.sh/agent/nous-research`## Placement & structure that works- SKILL.md at **repo root** — that's what `npx skills add` picks up.- Self-contained: the agent has zero other context. Include: - When to use (trigger) - Every command with examples - Output format (JSON field names) - Platform notes (e.g. "must run on Windows, else returns []") - Agent rules (never fabricate; risk is authoritative; writes not yet available)## Worked example (BootKeeper)`publieople/BootKeeper`: root SKILL.md documents `bootkeeper list/get/analyze/snapshot` (JSON out). CLI builds cross-platform; SKILL.md documents JSON shapesan agent parses directly.## Pitfalls- **Don't skip the CLI.** A doc-only SKILL.md gives the agent nothing to run — it would write its own fragile reimplementation each time.- **Don't bake MCP into core.** MCP is an adapter; the CLI is the contract.- **Verify install format**: if `npx skills add` fails frontmatter validation, fall back to git clone (see hermes-external-skill-install skill).

Frequently asked questions

Cross Agent Skill Publishing