Imported from onehorizonai/skills (AGENTS.md). Install upstream with npx skills add onehorizonai/skills. Copyright stays with the author.

Agent Guide

This repo contains one plugin. skills/ is the source of truth for skill instructions. App-specific manifests live at the repo root and are generated from plugin.json. Do not edit generated files directly.

Structure

skills/
  {skill-name}/
    SKILL.md
    agents/openai.yaml
    assets/

assets/                             # Shared plugin assets
hooks/                              # Shared hook scripts
scripts/
  build-manifests.mjs               # Regenerates app-specific manifests
  validate.mjs                      # Validates manifests and skill metadata
  validate-links.mjs                # Validates local refs and external URLs

plugin.json                         # Canonical plugin manifest
.mcp.json                           # Shared MCP server config
copilot-hooks.json                  # Shared hook manifest
.cursor-plugin/plugin.json          # Generated Cursor manifest
.codex-plugin/plugin.json           # Generated Codex manifest
.claude-plugin/marketplace.json     # Claude marketplace
.agents/plugins/marketplace.json    # Local Codex marketplace
.github/plugin/marketplace.json     # GitHub/Copilot marketplace

Working with skills

Each skill lives at skills/{skill-name}/SKILL.md with frontmatter and agent instructions.

SKILL.md format

---
name: {skill-name}          # Matches the folder name; no plugin prefix
description: {One sentence with trigger phrases like "prep my standup" or "list blockers".}
---

# Skill title

What this skill does, when to use it, and which nearby skill owns adjacent work.

## Instructions

Step-by-step instructions for the agent to follow when this skill is invoked.

Rules

• name must match the folder name exactly; do not add a plugin namespace prefix
• The slash command is built automatically: /{plugin-name}:{skill-name}
• Keep SKILL.md short because the full file loads into context on every invocation
• description controls when the agent picks up the skill; write it with concrete trigger phrases
• Edit skills in skills/ only
• Don't edit generated app manifests unless you're also updating scripts/build-manifests.mjs

Rebuild manifests

After editing plugin.json or marketplace metadata logic:

node ./scripts/build-manifests.mjs

This regenerates .cursor-plugin/plugin.json, .codex-plugin/plugin.json, and marketplace files from the root manifest.

Validate

node ./scripts/validate.mjs          # Manifest consistency, skill metadata, required files
node ./scripts/validate-links.mjs    # Local refs and user-facing URLs

Adding a skill

1. Create skills/{skill-name}/
2. Add SKILL.md
3. Add agents/openai.yaml
4. Add assets/logo.svg
5. Run node ./scripts/validate.mjs
6. Run node ./scripts/validate-links.mjs if you added references or URLs

Plugin manifest fields

{
  "name": "plugin-name",
  "description": "One sentence shown in the plugin manager.",
  "version": "1.0.0",
  "author": { "name": "Author Name" },
  "homepage": "https://example.com",
  "repository": "https://github.com/onehorizonai/skills",
  "license": "MIT",
  "skills": "./skills/",
  "mcpServers": "./.mcp.json",
  "hooks": "./copilot-hooks.json"
}

name sets the skill namespace. A skill in folder my-skill inside plugin my-plugin becomes /my-plugin:my-skill. plugin.json is canonical, so keep generated manifests aligned with it.