How is CueCrux different from agent observability dashboards?

Observability tools collect logs and traces, then draw dashboards over them. Useful, but a log is an assertion: whoever controls the pipeline controls the story, and nothing stops history being edited, sampled, or lost.

CueCrux differs in kind, not degree:

• Evidence, not assertions. Receipts are signed and hash-chained; verification works offline. An auditor does not have to trust your logging pipeline or our UI.
• Attribution, not estimates. Costs trace fleet to agent to verb to receipt. There is no sampling step where numbers become approximate.
• Controls, not just visibility. RCX policy refuses out-of-scope calls with signed refusal receipts, and revocation propagates with per-agent acks and a closing attestation.

Dashboards tell you what probably happened. CueCrux lets you prove what happened, and stop what should not happen next time.