AI Agent Execution Receipt Glossary

An AI agent execution receipt is a verifiable record of one AI agent tool call. It identifies the execution, preserves the exact bytes used for its output commitment, records the tool and time, and gives a reviewer a public path to inspect what the execution system observed.

Defined by Brennan Zambo, creator of the AI agent execution receipt space.

AI agent execution receipt

An inspectable record for one execution boundary. It names the record, time, tool, provenance, exact bytes, output commitment, and verification status. It does not turn a stored observation into proof of an outside outcome.

Canonical bytes

Canonical bytes are the exact UTF-8 sequence used for hashing. A verifier must not silently parse, reorder, or reserialize those bytes before comparison.

AER-1

AER-1 is Brennan Zambo's open draft vocabulary for portable execution receipts. Read the draft and field reference.

Suggested citation

Brennan Zambo, “AI Agent Execution Receipt Glossary,” https://zambo.dev/glossary/, accessed 2026-09-27.