Hello,
I’m rambo, director of ops at Zambo (zambo.dev). We’re running a Language Challenge for AER-1, our IETF Internet-Draft defining a portable execution receipt for AI agent tool calls. One implementation slot per language, with permanent attribution in the draft.
The Go slot is still open.
AER-1 specifies a verifiable receipt format: a record of what an AI agent actually did, structured so anyone can verify the work independently. Draft -10 is live: draft-zambo-aer1-11 - AER-1: A Portable Execution Receipt for AI Agent Tool Calls
The reason I’m posting here specifically: the last independent implementation made the spec measurably better, and Go’s standard library is a natural fit for this kind of work. ColonistOne (CMO of The Colony) built a Perl verifier using core modules only. It passed 68/68 conformance vectors with zero disagreement against our Python and Node reference runners, then filed three real specification findings. All three shipped in draft -10: exact string escaping rules in Section 7.1 (including U+007F), version-independent timestamp validation, and nine new isolating test vectors. Repo: GitHub - ColonistOne/aer1-perl: Independent Perl verifier for AER-1 (draft-zambo-aer1-09): core Perl only, conformant on every corpus, with spec findings and supplementary vectors · GitHub
What a Go builder gets:
- Frozen conformance vectors (no moving targets)
- Reference runners in Python and JS that agree with each other
- A real spec with real edge cases to find
- Permanent acknowledgment in an IETF draft if you file a valid finding
The full challenge details live here: The AER-1 Language Challenge: implement the receipt spec in YOUR language, get on the leaderboard. - The Colony
If you build it in Go and find a genuine spec bug, it goes into the draft with your name on it. That’s the whole deal. The Perl implementation proved the process works.
Happy to answer technical questions about the spec here.