An agent just acted on your behalf.
Can you prove what it did — and why it was allowed to?

This scripted scenario depicts a generic agent handling four consequential actions. Synthetic receipt records are evaluated against illustrative policy rules and linked in a browser-generated SHA-256 chain. Flip between Without assurance and With Summit, open a record, then verify the chain and try to tamper with it.

ILLUSTRATIVEEvery actor, record, policy result, and event on this page is synthetic. No agent, external tool, policy service, or system of record executes. Only the in-browser WebCrypto SHA-256 hash calculation is live.

Run the actual live, input-driven service proof →

Scenario

The script depicts an operations agent assigned request REQ-4821 on a synthetic account. The four steps below illustrate how a policy-gated workflow could be recorded; they are not executed actions.

Signal inagent action

Policy gatedeny by default

Decisionallow · escalate · block

Synthetic recordbrowser hash-linked

Verify & replayby anyone

Every scripted step renders a synthetic receipt record — click one to inspect it.

ACCEPTED — allowed by policy ESCALATED — handed to a named human BLOCKED — denied, out of scope

The reviewer's question

Same four actions, two answers. A reviewer months later asks what actually happened.

Reviewer asks	Without assurance	With Summit
Was each action captured as a record at all?	Scattered logs, short retention	One depicted receipt per scripted action
Who acted, and on whose authority?	A generic service account	Actor + authorizing policy on every receipt
Why was a consequential action allowed?	Inferred after the fact	Evidence and rule verdict, recorded up front
Could the record be altered afterward?	Anyone with log access	Tampering breaks the chain — try it below