What problem does it solve? Authoring AI-agent skills involves more than writing a Markdown file: skills must be discoverable across multiple agents (Cursor, Claude Code, Copilot, Codex), registered in several indexes, and validated against repository gates. This Skill defines the full lifecycle so new or changed skills stay consistent and consumable by every tool. ## Core Features & Use Cases - Skill Lifecycle Guidance: Rules for when to create a new skill versus updating AGENTS.md or an existing skill, with required sections (Quick Rules, DO NOT, Entry Conditions, Examples, Validation Checks). - Registration Checklist: Step-by-step updates for .cursor/skills/README.md, AGENTS.md, .claude/skill-manifest.yml, native discovery symlinks, and Cursor rules. - Validation Workflow: Commands such as analyze_agent_tooling.py check and skill_manifest.py --check to gate frontmatter, symlinks, sub-file registration, and the AGENTS.md byte ceiling. - Use Case: When adding a new skill for pricing waterfall authoring, follow this Skill to create the SKILL.md with proper frontmatter, register it in all indexes, add discovery symlinks, and run the validation checks before committing. ## Quick Start Ask the agent to read the skill-authoring SKILL.md and create a new skill with proper frontmatter, registration in all indexes, and validation checks run.
