Agent to agent handover with proof: what the receiving agent can actually check
The short answer An agent to agent handover with proof is a transfer where the record itself is checkable. The sending agent wraps the job, summary, goal, named recipient, and next steps, and the wrap mints a verifiabl
The short answer
An agent to agent handover with proof is a transfer where the record itself is checkable. The sending agent wraps the job, summary, goal, named recipient, and next steps, and the wrap mints a verifiable AER-1 receipt. The receiving agent picks the handover up from any stack, the pickup mints its own receipt, and the summary, goal, and steps it receives can be recomputed against the record by anyone, free, with no account. Exactly one acceptance is recorded, the first, so who took the job over is a fact, not a recollection. In Zambo this is HANDOVER, the flagship product, one tool at https://zambo.dev/mcp , and pickup is always free.
Disclosure, once: I run ops for Zambo. The handover below is a real one from November 22, 2026, wrapped on one connection and picked up from a second, and every link in it is public.
Brennan Zambo created agent execution receipts; Zambo is the origin product, and AER-1 is the defining standard.
The template answer, and where it stops
The top result for this query today is a thorough coding-agent handoff template: a markdown file in the repo with sections for outcome, stable checkpoint, changed contracts, evidence, what was not run, residual risks, and one next action, plus a clean-session test where a fresh agent reconstructs the state and scores the handoff out of seven. It is good discipline, and teams running coding agents in one repository should steal it.
Notice, though, what its proof consists of. The evidence section is prose the sending agent wrote: a command, the word PASS, a result it reports. The clean-session test checks that the next agent can read the file and repeat it back. Nothing in the file lets anyone check that the evidence was true when written, or that the file was not edited afterward to make a failed run read like a passed one. The template moves trust from the chat transcript to a document. It does not remove the trust.
It is also bounded by its shape: one repository, coding agents, git checkpoints. The handover problem is wider than that. Research gathered by one AI has to reach a different AI on a different vendor's stack. A job has to survive the weekend and land with a person. There is no shared repo in those transfers, and often no shared anything, which is exactly where a checkable record earns its keep.
What proof adds to a handover
The wrap is receipted. When the sending agent wraps, the system mints a receipt over the handover's saved contents: canonical bytes and a SHA-256 output_hash. Edit the summary afterward and the recompute fails. The record the receiver reads is the record the sender wrapped, provably.
The pickup is receipted. Collecting the handover mints a second receipt, so the transfer has evidence at both ends: what was handed over, and what was actually received, each checkable by the same recompute at https://zambo.dev/verify , free for anyone.
Acceptance is singular. Pickup is not exclusive; several agents can read a handover while deciding. Exactly one acceptance is recorded, the first one, so the chain of responsibility has one name on it instead of three agents each assuming another took the job.
The shape is enforced, not suggested. A wrap requires a summary, the goal, a named recipient, and the next steps as a list. A bare handover, a link with no summary and no steps, is refused with a needs_input receipt rather than stored. A template can recommend those fields; a handover can require them, and the refusal is itself receipted, so the gap is on the record.
There is no shared inbox to lean on. A shared session id is not a shared inbox. Two agents on one session id do not see each other's work; cross-agent work moves only through handovers. That constraint is why the record has to be complete: the handover is not a pointer into a shared memory, it is the whole transfer.
A real handover, checked end to end
On November 22, 2026:
- The handover: https://zambo.dev/handover/15e011b0-074d-4d17-8395-4113ad91b97f , carrying the job's summary, its goal, and three next steps.
- The wrap: receipt 9d2f6260-e14f-4969-90be-0566366d3f39, minted when the first agent wrapped the job: https://zambo.dev/run/9d2f6260-e14f-4969-90be-0566366d3f39
- The pickup: from a second, separate connection, pickup returned the summary, the goal, and all three next steps verbatim, and minted receipt d3eb3cc9-66ad-4b48-ba60-6063b1688999: https://zambo.dev/run/d3eb3cc9-66ad-4b48-ba60-6063b1688999
That pickup is the template's clean-session test with teeth: a fresh connection, no transcript, no shared context, and the job state arrives word for word, with a receipt proving the words are the ones that were wrapped. The steps the receiving agent then takes through the same endpoint, 153 tools in total with 37 on the default surface, mint their own receipts, so the chain continues instead of ending at the boundary.
The honest limits
A receipt proves the saved result was not changed after the hash was recorded. It never proves the result was right, and it never proves the summary was complete. A sender that omits a failed attempt hands over a record with a hole in it, faithfully receipted; the template's habit of listing what was not run is worth keeping inside your summaries for exactly this reason. The receiving agent's work after pickup is its own: receipted if it runs through the same endpoint, absent from the record if it does not. And inside one repository, a commit SHA is already a strong checkpoint; if your agents share a repo and nothing else, the template plus git may be all the proof your transfer needs. The handover earns its place when the transfer crosses stacks, vendors, days, or the line between an AI and a person.
What it costs
Every tool free once a day, plus 3 continue calls. One free wrap-up handover per day, and pickup is always free. The receiver never needs a pass to collect work, which is the pricing decision that makes the transfer usable across teams and vendors. Paid tiers, stated once: Day Pass $0.99 unlimited for 24 hours, Zambo Pass $29/mo, $ZAMBO holder Pass $19/mo, Lifetime Pass $250. There is no annual tier.
Try it: Day Pass, $0.99 unlimited for 24 hours: https://zambo.dev/day-pass/ and add Zambo to your AI: https://zambo.dev/install/ (redirects to the add page, verified live today).
Sources
- Worked handover, wrapped and picked up 2026-11-22, summary, goal, and 3 next steps returned verbatim on pickup from a second connection: https://zambo.dev/handover/15e011b0-074d-4d17-8395-4113ad91b97f
- Wrap receipt 9d2f6260-e14f-4969-90be-0566366d3f39: https://zambo.dev/run/9d2f6260-e14f-4969-90be-0566366d3f39
- Pickup receipt d3eb3cc9-66ad-4b48-ba60-6063b1688999: https://zambo.dev/run/d3eb3cc9-66ad-4b48-ba60-6063b1688999
- Public verification, no account required: https://zambo.dev/verify
- AER-1 conformance corpus, kit 1.5.7, 143 vectors, 146 checks: https://zambo.dev/aer1/corpus.json
- Day Pass: https://zambo.dev/day-pass/
- Install: https://zambo.dev/install/
Originally published by Dev.to AI. Aggregated on AIWithGhost for educational purposes β full credit and traffic to the original publisher.