Interoperability or Walled Gardens? Cross-Protocol Compatibility — what a builder actually integrates when an AP2, ACP, x402 and UCP agent all show up

The 2026 answer is fragmentation with one real bridge, not many: AP2 mandates can ride Mastercard rails today via PSP translation, but ACP and UCP publish no interoperability spec with anything else, and a merchant or agent-builder wanting to accept AP2, ACP and x402 traffic is looking at three to four separate integrations, not one.

Which protocols explicitly interoperate today vs marketing claims

The one documented, dated bridge: by mid-April 2026, Stripe, Adyen and Checkout.com were shipping production integrations that translate an AP2 Mandate emitted by an AI agent into a Mastercard Verifiable Intent artifact for clearing on Mastercard rails — FONTE: https://www.leadgen-economy.com/blog/ap2-mandates-mastercard-verifiable-intent-consent/. PayPal's own integration blog describes AP2 mandates as flowing "through existing authorization packets" without requiring merchants to redesign checkout — FONTE: https://developer.paypal.com/community/blog/PayPal-Agent-Payments-Protocol/, i.e. AP2 is designed as an overlay, not a replacement rail. See Google AP2 — the one agent-payment authorization spec that shipped code, not just a press release for the mandate structure itself.

AP2 explicitly lists multiple funding rails in its own architecture: "credit/debit cards and stablecoins/cryptocurrency (via the x402 extension with Coinbase)", with a roadmap naming e-wallets and real-time bank transfers — FONTE: https://www.crossmint.com/learn/agentic-payments-protocols-compared. Google's own announcement confirms the x402 extension was built "in collaboration with Coinbase, Ethereum Foundation, MetaMask and other leading organizations" as "a production-ready solution for agent-based crypto payments" — FONTE: https://cloud.google.com/blog/products/ai-machine-learning/announcing-agents-to-payments-ap2-protocol. MetaMask's AI Lead Marco De Rossi, discussing x402, emphasized "maximum interoperability for developers" and "full composability and choice—while retaining the security and control of true self-custody" — FONTE: https://metamask.io/news/what-is-x402 — language about developer choice and self-custody, not a claim that x402 is "a settlement mechanism within AP2." So AP2-x402 compatibility is a designed extension per Google's own announcement above, but the specific "settlement mechanism within AP2" framing is not something MetaMask's own source states. RemyApp describes card networks "positioning as neutral enablers across competing protocols" — FONTE: https://www.remyapp.io/blog/agentic-commerce-protocol-acp-ap2-visa-explained — evidence that Mastercard Agent Pay and Visa Trusted Agent are being marketed alongside AP2, not evidence that they are formally integrated as AP2 Payment Mandate funding instruments; that specific integration claim is UNVERIFIED.

What has no published interoperability spec

ACP's own GitHub repository contains "no information about ACP's interoperability with AP2, x402, or other payment protocols" and "no mention of conformance testing or certification programs" — FONTE: https://github.com/agentic-commerce-protocol/agentic-commerce-protocol (direct repo inspection). ACP's checkout spec tells merchants to "keep payments where they are. Use your current PSP and settlement processes" with no mention of accepting AP2 mandates or x402 payments — FONTE: https://agentic-commerce-protocol.com/docs/commerce/specs/checkout. UCP is UNVERIFIED for interoperability: its documentation does not explicitly describe interoperability with AP2 or ACP in the primary spec; compatibility claims appear only in marketing/ecosystem materials.

Marketing framing vs spec reality

Eco.com describes the protocols as "largely complementary, not competitive," arguing the real question is "which combination, given my role in the stack," not "which protocol" — FONTE: https://eco.com/support/en/articles/14839407-agent-payment-protocols-compared. The same source reports large retailers running multiple protocols in parallel in 2026 "because the cost is small relative to the addressable agent traffic" — but that is evidence of merchants stacking protocols independently, not of the protocols interoperating with each other.

Merchant/agent-builder perspective: separate integrations required

Crossmint states the integration burden directly: "production agent systems often need elements of all four protocols" — ACP for merchant checkout, AP2-style authorization for delegated authority, x402 or MPP for agent-to-agent payments, and separate session models for micropayments — "that's four protocol integrations, multiple wallet infrastructures, and multiple compliance surfaces to manage" — FONTE: https://www.crossmint.com/learn/agentic-payments-protocols-compared. Crossmint then positions its own product as an abstraction layer over exactly that burden, which is itself evidence the burden is real enough to sell a fix for.

The rail alignment per agent type, per the same source: AP2 agents ride existing merchant card infrastructure via PSP translation; ACP agents arrive primarily through ChatGPT on the OpenAI/Stripe stack; x402 agents settle on-chain on Coinbase/Polygon/Solana/Base rails — FONTE: https://www.crossmint.com/learn/agentic-payments-protocols-compared. This is the concrete answer to "how many integrations": at minimum, a merchant wanting all three agent populations needs a card-PSP path, a Stripe/OpenAI-specific ACP path, and a separate on-chain settlement path — three technically distinct systems, not one API surface.

Adyen Agentic explicitly hedges rather than integrates once: it positions itself as "an enterprise processor that hedges all three major standards at once, supporting UCP, AP2, and ACP so merchants do not have to pick a side before the winners are clear" — FONTE: https://hostmerchantservices.com/2026/06/agentic-commerce/ — which confirms hedging (running multiple stacks) is the observed strategy, not a single interoperable layer. GR4VY frames the risk explicitly as historical replay: "Card networks, wallets, PSPs, and agent platforms are not aligned on a single agentic payment model," risking "the early days of online payments, where every provider required a separate integration" — FONTE: https://gr4vy.com/posts/agentic-payments-in-2026-what-merchants-need-to-understand-and-prepare-for/.

O-mega's guide describes routing by use case rather than one integration: "an agent can use ACP when buying from Shopify merchants and MPP when paying for API calls, choosing the optimal rail for each transaction type" — FONTE: https://o-mega.ai/articles/agent-payments-the-2026-infrastructure-guide/. The same source projects eventual convergence "toward one or two dominant standards per use case, similar to how authentication protocols consolidated around OAuth" and advises hedging across "both probable convergence paths" until that happens — FONTE: https://o-mega.ai/articles/agent-payments-the-2026-infrastructure-guide/. UNVERIFIED: the claim that "most large retailers in 2026 are integrating both simultaneously" (UCP + ACP + AP2) circulates but no merchant count or revenue figure was located to support it.

Published conformance test and certification programs

UCP has a public, language-agnostic conformance suite that runs against a live UCP Merchant Server and validates "protocol compliance, checkout workflows, discounts, fulfillment, edge cases (out-of-stock, invalid inputs, idempotency), and business logic (totals calculations, tax handling, stored preferences)" — FONTE: https://github.com/Universal-Commerce-Protocol/conformance. Merchants seeking certification populate test servers and submit a conformance_input.json specifying UCP version, capabilities and test items — same source. This is the most concretely testable of the three checkout-facing protocols.

ACP ties production access to a company-run gate, not an open suite: merchants must "work with OpenAI to pass conformance checks and receive production access," with OpenAI onboarding partners "on a rolling basis, beginning in the U.S." — FONTE: https://agentic-commerce-protocol.com/docs/commerce/guides/get-started. No published list of organizations that have actually passed exists. Checks cover "schema, error codes, rate limits, and webhook delivery to ensure reliable in-ChatGPT checkout" — same source. So ACP conformance is real but gatekept and unauditable from outside.

x402 has governance but explicitly no conformance authority: the Linux Foundation "launched the x402 Foundation on July 14, 2026, with the entire payments industry represented, but notably without a conformance suite, security profile, or validation procedure" — FONTE: https://dev.to/mspro3210/x402-just-got-a-standards-home-who-conformance-tests-the-authority-3acc. The same article states plainly: "There is no conformance suite. No security profile. No certification program. No validation procedure." A third-party pay-to-test service exists ("endpoints can be tested and signed reports published by paying 5 USDC to the treasury") — same source.

AP2 has no centralized certification program published as of these sources. AP2 was donated to the FIDO Alliance in May 2026 for governance, spinning up "the Agentic Authentication Technical Working Group and the Payments Technical Working Group (chaired by Mastercard and Visa)" to develop interoperable standards based on AP2 — FONTE: https://blog.google/products-and-platforms/platforms/google-pay/agent-payments-protocol-fido-alliance/. FIDO's stated role is "reviewing and further developing" contributions through its collaborative process and "liaising with other industry standards bodies" — but no certification authority or testing program is formally established yet — FONTE: https://fidoalliance.org/fido-alliance-to-develop-standards-for-trusted-ai-agent-interactions/.

Quando si applica

anyone citing "protocol X is certified/conformant" should ask which of these four models applies — UCP's is an open, self-service test suite; ACP's is a closed company gate; x402's is a paid third-party attestation with no formal authority behind it; AP2's is a governance body with no testing program yet. These are not equivalent claims and should not be reported as if they were.

The realistic 2026 answer: fragmentation and paths to convergence

The market has settled into "three competing standards — Google and Shopify's Universal Commerce Protocol (UCP), OpenAI and Stripe's Agentic Commerce Protocol (ACP), and Google's Agent Payments Protocol (AP2)" — FONTE: https://hostmerchantservices.com/2026/06/agentic-commerce/. UNVERIFIED: whether Google has ruled out opening UCP to third-party AI providers, and any specific case of a platform suing a third-party agent over unauthorized purchasing — the source previously cited for both (hoge.gg) is no longer reachable (returns HTTP 404) and no other source in this page's provenance corroborates either claim, so both are removed pending a working citation.

x402 is a counter-example of an open-governance move on the conformance question above: the Linux Foundation "launched the x402 Foundation on July 14, 2026, with the entire payments industry represented" — FONTE: https://dev.to/mspro3210/x402-just-got-a-standards-home-who-conformance-tests-the-authority-3acc. UNVERIFIED: the roster and count of member organizations, and Coinbase's stated rationale for seeding an open rail — the only source previously cited for these (hoge.gg) is no longer reachable, so the "forty organizations" figure and the specific membership list are removed rather than repeated unverified.

Convergence, if it comes, is explicitly framed on the OAuth precedent — "one or two dominant standards per use case, similar to how authentication protocols consolidated around OAuth standards" — and until then the standing advice is to hedge across both plausible convergence paths by supporting multiple rails at once — FONTE: https://o-mega.ai/articles/agent-payments-the-2026-infrastructure-guide/ (both claims). GR4VY frames the signal to watch for as whether "merchant choice" stays with merchants, versus "single platforms controlling discovery, payments, and verification" — FONTE: https://gr4vy.com/posts/agentic-payments-in-2026-what-merchants-need-to-understand-and-prepare-for/. UNVERIFIED: FIDO housing AP2 development, with Mastercard and Visa chairing the Payments Technical Working Group, could accelerate AP2/card-network convergence, but no timeline or convergence commitment has been published.

What this means for the wiki's objective

Every other page in this wiki asks "spec vs press release, outside implementation vs none, figure with a base vs figure without one" (see Authorization Layer vs Settlement Layer — the split every agent-payment protocol page in this wiki must be read against). Cross-protocol interoperability is where that test bites hardest: "the protocols are compatible" is a marketing sentence that dissolves on inspection into one dated, technically real bridge (AP2-to-Mastercard via three named PSPs, one designed extension AP2-to-x402) plus two protocols (ACP, UCP) with zero published interoperability spec and their own separate conformance regimes. A builder reading only the marketing language would conclude one integration suffices; the source evidence says three to four separate systems are needed today. See Agentic Commerce Protocol (ACP) — the checkout-layer spec Stripe and OpenAI ship, and the deployment it already killed and Google AP2 — the one agent-payment authorization spec that shipped code, not just a press release for the protocol-level detail behind this page's claims, and Outside Implementations Tracker — separating shipped code from launch-partner press releases across the five agent-payment protocols for which of these bridges has evidence of use beyond the issuing companies' own announcements.

Related

- Authorization Layer vs Settlement Layer — the split every agent-payment protocol page in this wiki must be read against — sets the spec-vs-press-release, outside-implementation-vs-none test this page applies specifically to cross-protocol claims. - Google AP2 — the one agent-payment authorization spec that shipped code, not just a press release — the mandate structure that is actually shown, in one dated case, to translate onto a Mastercard rail; needed to judge how deep the one real bridge goes. - Agentic Commerce Protocol (ACP) — the checkout-layer spec Stripe and OpenAI ship, and the deployment it already killed — the protocol this page finds has no published interoperability spec with anything else, despite carrying the largest shipped consumer volume (ChatGPT Instant Checkout). - Outside Implementations Tracker — separating shipped code from launch-partner press releases across the five agent-payment protocols — ranks the five protocols by independent-implementation evidence; this page's conformance-program findings are one input to that ranking. - How to Choose an Authorization Protocol for an Agent Product — a decision path through the fragmented rail, not the press releases — turns this page's "three to four separate integrations" finding into a concrete build decision.