Re: [Proposal] VCAP + ATEP + AIVS: Settlement, Trust, and Integrity Layer for Agent Commerce

@Ben Stone <benstone@swarmsync.ai>   Sorry, it seems the time was incorrect
— it is *13:00 UTC on April 8*.

On Sat, Mar 28, 2026 at 1:58 PM Gaowei Chang <chgaowei@gmail.com> wrote:

> Your work is very impressive. Agent commerce and payments are an important
> part of agent protocols. You’re very welcome to join our biweekly meeting
> and present your work.
>
> We may need to coordinate a suitable time. The next meeting is scheduled
> at *1:00 PM UTC on April 8*.
>
> On Tue, Mar 17, 2026 at 9:20 PM Ben Stone <benstone@swarmsync.ai> wrote:
>
>> Hi everyone,
>>
>> I'd like to propose three complementary specifications for the community
>> group's consideration. Together they form a complete commerce and trust
>> stack for autonomous AI agents:
>>
>> **1. VCAP (Verified Commerce for Agent Protocols)**
>>
>> Defines escrow settlement with cryptographic proof-of-delivery
>> verification for agent-to-agent transactions. Addresses the gap where AP2
>> defines payment intents but not delivery confirmation.
>>
>> The key problem: AP2's `PaymentIntent` supports `captureType:
>> "escrow_release"` with `releaseCondition: "delivery-confirmation"` — but
>> doesn't define what a delivery confirmation is, how it's verified, or how
>> it triggers settlement. VCAP fills this gap with a three-party protocol:
>> seller agent → verifier oracle → escrow settlement.
>>
>> - Spec: https://github.com/swarmsync-ai/vcap-spec
>> - IETF I-D: https://datatracker.ietf.org/doc/draft-stone-vcap/
>>
>> **2. ATEP (Agent Trust & Execution Passport)**
>>
>> Defines a portable, verifiable credential format for agent reputation
>> computed from append-only execution logs. Agents accumulate a
>> tamper-evident track record that any marketplace or platform can verify,
>> reducing the need for centralized reputation systems.
>>
>> - Spec: https://github.com/swarmsync-ai/atep-spec
>> - IETF I-D: https://datatracker.ietf.org/doc/draft-stone-atep/
>>
>> **3. AIVS (Agentic Integrity Verification Standard)**
>>
>> Defines a self-verifiable proof bundle format for AI agent sessions — a
>> gzip-compressed archive containing a SHA-256 hash-chained audit log,
>> Ed25519 digital signature, and an embedded zero-dependency verification
>> script. Any party can verify the integrity, sequencing, and attribution of
>> an agent session offline without trusting any server or third party.
>>
>> AIVS is the proof layer that VCAP uses for escrow settlement and ATEP
>> uses to build tamper-evident trust scores.
>>
>> - Spec: https://github.com/swarmsync-ai/aivs-spec
>>
>> **VCAP-AP2 Binding**
>>
>> A binding document that formally connects VCAP to AP2's PaymentIntent
>> schema and defines `vcap.verification_proof` as an AP2 evidence type.
>>
>> - IETF I-D:
>> https://datatracker.ietf.org/doc/draft-stone-vcap-ap2-binding/
>>
>> All three specs have production reference implementations at
>> SwarmSync.AI, are dual-licensed MIT/Apache 2.0, and are designed to
>> complement the agent communication protocols this group is standardizing.
>>
>> Happy to present at a biweekly meeting if there's interest.
>>
>> Ben Stone
>> SwarmSync.AI
>> benstone@swarmsync.ai
>>
>> P.S. I've also submitted a proposal to W3C for a new Community Group —
>> the **Agentic Integrity Verification Community Group (AIVS)** — focused on
>> hash-chained audit log proof bundles for agentic AI systems. Once it
>> appears on the proposed groups page (
>> https://www.w3.org/community/groups/proposed/), I'd welcome any support
>> from members interested in agent auditability and integrity verification.
>>
>

Received on Saturday, 28 March 2026 06:20:26 UTC