coding-ethos First Run: Policy Seams, Agent Hooks, and Remediation Evidence

The first useful test for an AI coding guardrail is not whether it can print a rule. It is whether an operator can replay the decision after the agent has been stopped. coding-ethos treats that as a s