Expose a stable machine-readable skill catalog

[Scriptception]

Problem

Skill metadata is validated internally, but downstream tools must import Python internals or scrape command output to discover versions, categories, compatibility, source references, and evaluation status.

User outcome

Tools and users can inspect the official catalog through a stable documented data contract.

Scope

• Define a versioned catalog schema generated from canonical metadata.
• Include identity, version, category, description, supported agents, content digest, source, and deprecation state.
• Expose JSON through the CLI and release artifacts.
• Keep generated output freshness-tested.

Non-goals

• Hosting a remote service.
• Adding organization-specific approval or install data to the public catalog.

Acceptance criteria

• Catalog output is deterministic and schema-versioned.
• Every bundled skill appears exactly once with a verifiable digest.
• Consumers can filter without parsing human-formatted CLI text.
• Schema changes follow documented compatibility rules.

Validation

• Validate output against a committed JSON Schema.
• Compare catalog digests with installed stamp calculations.

Relationships

• Supports organization bundles and authenticated catalogs.
• Feeds public documentation without replacing canonical metadata.