skill-authoring-workflow

The skill-authoring-workflow acts as the primary quality control mechanism for the Product Manager Skills repository. It provides a structured, dogfood-first methodology for PMs to transform half-baked prompt dumps, rough workshop notes, or experimental frameworks into professional-grade assets that adhere to strict repository standards. By utilizing native scripts such as find-a-skill.sh, add-a-skill.sh, and test-a-skill.sh, this workflow eliminates 'looks good to me' guesswork and ensures every new contribution is immediately deployable and discoverable. It is designed to minimize friction for contributors while maximizing the reliability of the resulting AI agents.

• Streamlines the transition from unstructured ideas to fully formatted skills/SKILL.md assets.

• Enforces strict validation gates for frontmatter, metadata limits (name ≤ 64, description ≤ 200 chars), and section ordering.

• Facilitates content generation using guided wizards or context-first generators based on source materials.

• Automates repository integration tasks, including README table updates, skill count increments, and cross-reference resolution.

• Provides a robust testing suite for smoke testing, metadata auditing, and trigger-word verification.

• Includes support for packaging skills into distributable ZIPs for various agent environments like Claude Desktop, Codex, and enterprise-grade CLI tools.

• Designed for PMs, developers, and AI engineers who need to scale their internal toolkits without technical debt.

• Follows the workshop-facilitation protocol when acting as an interactive advisor.

• Mandatory usage of the check-skill-metadata.py script is required to satisfy 'Definition of Done' requirements.

• Ideal for converting legacy documentation into executable agent logic.

• Avoids common pitfalls like bloated descriptions, ambiguous trigger language, and broken dependency links.

• Requires consistent usage of standard CLI tools to avoid 'vibes-based' development and ensure predictable output quality.