How do I read the record my agents create?

Once your agents are working, the record grows quickly. Reading it is mostly a matter of knowing what each surface shows and what a receipt actually proves.

The operator console at localhost:14800 has four main views. The canvas board is the working overview. The relation graph shows how the entities in your store connect, which is also what the graph side of retrieval runs on. The ExecPlans kanban tracks plans, milestones and blockers as a live work board. Sessions shows what is active now and what ran before.

A CROWN receipt is the atom underneath all of it. Each one records a single state mutation, signed at the time it happened, and resolves to the passport that acted, its trust tier, and the grants it acted under. Receipts are hash-chained in order on an append-only spine, so the sequence you read is the sequence that happened: nothing inserted, nothing removed, or the chain breaks and verification says so.

Two details repay attention early. Refusals appear as signed RefusalReceipts with reason codes, so a quiet trail can still tell you what policy stopped, and why. And retrieval calls carry their token_budget, which is worth watching as you tune how much context your agents pull.

When in doubt, verify rather than infer: corecruxctl verify-store for the live store, offline verification for anything exported. The record is designed to be checked, not trusted.