Agent Payments Protocol (AP2) by Google (standardisation moved to the FIDO Alliance)

# Fix list: Agent Payments Protocol (AP2)

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/ap2, the October 2026 research run, assessed 1 October 2026. Grade C, 55.3 out of 100.

This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public.

For a coding agent working on Agent Payments Protocol (AP2): work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published.

## 1. Reliability, 31 out of 100, up to 13.8 more on the total

Why it scored 31: Protocol reading, split as reference implementations 30, live deployments 25, spec stability 25 and test vectors 20. A Python SDK for SD-JWT mandates, constraints and receipts, with sample roles in Python, Go and Android, installed from git because there is no PyPI package (18 of 30). Neither the v0.2 release post nor the docs name a production deployment, and we found none elsewhere (0 of 25). v0.2 on 28 April 2026 replaced the Cart and Intent Mandates with open and closed Checkout and Payment Mandates, the protocol is pre-1.0, and ap2-protocol.org/specification/ still serves the v0.1 text while v0.2 lives at /ap2/specification/ (8 of 25). Nine unit-test files in the SDK, but CI runs only a linter on pull requests, and issue #265 proposing conformance vectors is still open (5 of 20).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):

Hosted APIs, MCP servers, models and platforms.

- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).
- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.
- 15, rate limits documented with numbers.
- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.
- 10, an SLA published for any paid tier.
- 10, the surface agents use is generally available, not beta or preview.

Local packages, SDKs, frameworks and stdio MCP servers.

- 20, installs from an official package with supported runtimes stated.
- 25, a public CI and test suite, passing on the default branch.
- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).
- 15, semver discipline and breaking changes called out in a changelog.
- 15, version 1.0 or later, or declared stable.

Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.

## 2. Agent ergonomics, 51 out of 100, up to 8 more on the total

Why it scored 51: Protocol reading. An implementer has to hold seven documents and SD-JWT key-binding chains before a first purchase (12 of 25). Selective disclosure means the agent presents only the constraints a verifier needs (10 of 20). Rejection receipts tell the agent a mandate failed, without documented reason codes (10 of 20). The double-spend rule (no second open mandate until a rejection receipt arrives) and the binding of each Payment Mandate to one checkout hash make retries safe by design (14 of 20). One SDK, Python only, from git; the FAQ says an SDK and MCP server are being built (5 of 15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):

- 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries).
- 20, pagination, filtering and output-size controls.
- 20, actionable, documented error responses, codes and messages an agent can recover from.
- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.
- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.

Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs.

## 3. Maintenance & community, 19 out of 100, up to 7.1 more on the total

Why it scored 19: Release v0.2.0 on 28 April 2026, 156 days before this check (10). No release or commit on main since 29 April (0). 50 open issues and 69 open pull requests, Dependabot bumps for `cryptography` left open, and issue #294 (10 July) on docs pointing at a missing spec file unfixed (3 of 25). The Python SDK isn't on PyPI and no other language has one (3 of 15). `cryptography` is pinned at 46.0.5 with newer releases waiting in open PRs, and CI is lint only (3 of 10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):

- 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older.
- 20, at least three releases or dated changelog entries in the last 90 days.
- 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15.
- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).
- 10, package health, current dependencies and CI.

Models are read for deprecation notice periods and model churn rather than release counts.

## 4. Payments & pricing, 60 out of 100, up to 5 more on the total

Why it scored 60: Protocol reading. Human-not-present mode lets an agent pay within user-signed open mandates without a person at the moment of purchase, and the repository has x402 samples for both modes, but a person must sign first and there is no live rail to pay on (20 of 40). The protocol defines no fees (20). Free to implement, nothing to buy (20). An agent can't obtain a mandate or a credential provider on its own (0).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):

The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).

- 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which.
- 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login.
- 20, a free tier or trial that doesn't need a card.
- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).

Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied.

Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol.

## 5. Schema & documentation, 74 out of 100, up to 4.2 more on the total

Why it scored 74: JSON Schemas for the six mandate and receipt types plus Pydantic models; AP2 has no HTTP API of its own, so there is no OpenAPI file to expect (20 of 25). Markdown sources in the repository, and the MkDocs build publishes ap2-protocol.org/llms.txt with ten links and an llms-full.txt (10). Seven spec documents, 2,469 lines, set out roles, who verifies what, both modes and the dispute evidence (16 of 20). Typed constraints such as `payment.amount_range`, `payment.budget` and recurrence with `max_occurrences` (12 of 15). Decoded mandate and disclosure examples throughout, and receipts that accept or reject, but no table of error codes (10 of 15). Mandate schemas are versioned by a `vct` suffix, but CHANGELOG.md has two one-line entries and the site serves two contradictory spec versions (6 of 15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):

APIs and MCP servers.

- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).
- 10, llms.txt or Markdown docs served for agents.
- 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference.
- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.
- 0 to 15, examples and documented error responses.
- 15, versioning and a public changelog.

Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference.

## 6. Transparency & trust, 56 out of 100, up to 3.9 more on the total

Made of editorial 55, provenance 57.

Why it scored 56: Apache-2.0, all source public (30). The spec has privacy rules (present only needed disclosures), but the site has no privacy policy or terms (10 of 30). v0.2 replaced the mandate types without a migration note we could find, and the v0.1 page is still live with no notice on it (5 of 20). A spec with no runtime of its own; the samples call Gemini with the user's Google key, which the README says (10 of 20).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):

- 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms.
- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).
- 0 to 20, a deprecation policy or notices with dates.
- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).

The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.

Provenance checks not met in full (half of this category, computed from checked facts):

- Domain age: ap2-protocol.org, registered 2025-09-15 (1 year) (3 of 15)
- Status page: not found (0 of 10)
- security.txt: not found (0 of 10)

## 7. Security & auth, 84 out of 100, up to 2.8 more on the total

Why it scored 84: User-signed SD-JWT mandates bound to the agent's key through `cnf`, a short `exp` recommended, and each closed mandate bound to a merchant-signed checkout by hash. We found no way to revoke an open mandate before it expires (24 of 30). Human-present purchases need the user's signature on a Trusted Surface, and open mandates cap amount ranges, budgets, recurrences, merchants and items (20). The threat model assumes prompt injection can't be prevented, treats every LLM as a potential attacker and bounds the damage with constraint checks at verification (15). Signed Checkout and Payment Receipts go to the agent, credential provider and network, and the mandates are designed as dispute evidence (13 of 15). SECURITY.md routes reports to Google's g.co/vulnz with a five-working-day response and GitHub advisories; no security.txt on ap2-protocol.org (12 of 20).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-security):

- 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option.
- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.
- 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10.
- 0 to 15, audit logs or per-call visibility for the operator.
- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.

Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing.

## What we couldn't check

What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it.

- Whether any credential provider or network runs AP2 in production; we found no primary source
- Whether the FIDO working groups will publish the spec themselves and on what timeline
- Whether open mandates can be revoked before expiry; the v0.2 docs don't say
- unchecked: arxiv 2601.22569 and 2609.00060, carried over from the listing's details

## Weaknesses

- No production deployment named by Google or found elsewhere
- No commit on main since 29 April 2026, with 50 open issues and 69 open pull requests
- The v0.1 spec page is still live and contradicts v0.2
- Python SDK only, installed from git, and no conformance vectors
- No documented way to revoke an open mandate before it expires

## What costs an agent a turn today

The notes we give agents before they call it. Each one is a workaround an agent shouldn't need.

- Read /ap2/specification/ for v0.2; /specification/ is the old v0.1 text
- Ask the user for open mandates with the shortest expiry that fits the task and a budget constraint
- Don't present a second open mandate until you hold a rejection receipt for the first
- Present only the disclosures the verifier needs
- Install the SDK from git; there is no PyPI package

## What the review panel asked for

- Document open-mandate revocation
- mandate revocation
- retire the v0.1 page

## When it's done

Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.