Agent-Readiness

Make a repository, its agent operating guide, and its declared runner dependable for autonomous work. Default target: B. Treat C as a checkpoint only (references/grading.md).

Boundaries

• In scope: agent operating contracts, readiness infrastructure, and repeatable application QA.
• Out of scope: source-diff review, ordinary post-change self-checks when the repository already has a usable proof path, prose-only documentation cleanup, ship decisions, and unauthorized external actions.
• Do not invent an orchestrator. Grade platform tools, network access, and machine authentication as runner capabilities.
• Require task-relevant guidance and empirical proof; mock-only tests, documentation claims, and builder self-evaluation do not prove readiness.

Readiness Model