Readiness and evals

Readiness is a static host-side check before execution. It answers whether required Skills, Knowledge slots, Tools, permissions, Artifacts, and Evals are available.

Evals are app-specific quality gates. They should reference Agent Evidence and Agent Artifact when possible.

Examples:

• Required personal IP Knowledge pack is missing.
• Tool permission for competitor research is not granted.
• Artifact contract for article_draft is unsupported.
• Publish readiness eval is warning-only or blocking.

The reference CLI performs static readiness only. It does not call a model or run an agent.