What problem does it solve? Coordinating multiple AI agent roles (Brain, Worker, Maintainer, Broker) in the SourceA workspace creates confusion about who judges ships, who edits code, and when progress claims are valid. This Skill defines the Brain role's decision boundaries, acceptance checklists, and routing rules so verdicts are based on disk state and script receipts rather than chat memory. ## Core Features & Use Cases - Ship Judging: Apply the ASF JUDGE checklist (disk SSOT match, zero critical bugs, validator PASS, founder law compliance) and emit ACCEPT or REJECT verdicts with a criterion table. - Hygiene Enforcement: Run registry hygiene and honest-status scripts before claiming progress, quoting Valid YES, receipts, or PARTIAL from script output instead of registry flags. - Work Routing: Direct hub UI and API changes to the Maintainer, INBOX turns to the Worker, and research requests to the Agent Hub intake. - Use Case: When a maintainer claims a hub fix is shipped, use this Skill to verify the disk state matches the order, run the validators, and issue a structured ASF JUDGE verdict. ## Quick Start Ask the agent to judge the latest maintainer ship using the ASF JUDGE checklist and quote the validator receipts.
