What problem does it solve?
Free-form SKILL.md files are human-readable but hard for agents to search, reason over, and audit. This Skill normalizes them into a structured Scheduling-Structural-Logical (SSL) JSON representation with typed layers, restricted enums, and graph validation.
Core Features & Use Cases
- Four-pass extraction pipeline: Extracts scheduling metadata, decomposes execution flow into scenes, expands scenes into atomic logic steps, then validates the result.
- Restricted vocabularies: Enforces fixed enums for scene types, action types, and resource scopes so outputs stay consistent and machine-comparable.
- Validation with bounded retries: Checks JSON syntax, enum membership, unique identifiers, pointer integrity, and graph reachability, retrying failed generations up to a bounded budget.
- Use Case: Given a repository of agent skill definitions, run the normalizer on each SKILL.md to produce searchable ssl.json files plus a per-artifact validation report for auditing.
Quick Start
Normalize the SKILL.md at .github/skills/ssl/SKILL.md into an SSL JSON representation and show me the validation report.