recipe-agent-team-operational-audit

You can perform a read-only audit of agent-team state to inspect run history, messages, and events without modifying state. This non-destructive approach ensures operational inspection and recovery assessment remain safe.

A read-only agent-team audit provides non-destructive visibility into runs, messages, events, and inboxes. It is used for operational inspection, runtime debugging, trace review, and recovery assessment without dispatching workers or mutating data.

To export the CLI schema for your agent-team deployment, run a read-only audit against your local state. This allows you to extract the contract and schema for tooling and documentation without altering active runs.

Yes, you can debug unresolved agent-team inboxes without triggering new runs by using a read-only audit. This inspect-only method identifies state gaps and stale tasks while strictly avoiding actions that create runs or dispatch workers.

Yes, state inspection requires the agent-team binary and the recipe's declared dependencies on agent-team-shared and agent-team-ops. These components are necessary to execute read-only actions and audit state without mutations.

Your agent-team audit needs to avoid mutating state to prevent unintended side effects during operational inspection. Read-only diagnostics ensure you can identify issues, trace events, and assess recovery safely without creating runs or modifying data.