WhatsApp Business Platform (Cloud API)

# Fix list: WhatsApp Business Platform (Cloud API)

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/whatsapp-cloud-api, the October 2026 research run, assessed 8 October 2026. Grade B, 67.1 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 WhatsApp Business Platform (Cloud API): 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. Payments & pricing, 40 out of 100, up to 7.5 more on the total

Why it scored 40: No x402, MPP or L402 (0). Per-message pricing is public without a login, as CSV and PDF rate cards in 16 currencies and an interactive table on whatsappbusiness.com. We couldn't open the rate files and scored the line on the public pages that list them (20). A test WhatsApp Business account and number send template messages with no payment method (20). A person has to create a Meta account, a developer app and a system user in a browser (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.

## 2. Security & auth, 67 out of 100, up to 5.8 more on the total

Why it scored 67: OAuth access tokens with permissions that separate messaging from account management, revocable, with a chosen expiry on system user tokens. The documented `debug_token` call passes the token to inspect as a query parameter, which costs 10 (20). System users can hold partial, view-only access to one WhatsApp Business account, and Meta reviews templates before they can be sent. The API has no approval step before a send (13). Inbound user messages are untrusted content. Meta's MCP documentation warns about prompt injection from webhook payloads and advises against write scopes for agents that read untrusted input. The Cloud API docs carry no such guidance (8). No audit log of API calls was found. Per-message status webhooks, `fbtrace_id`, and messaging and pricing analytics give partial visibility (7). security.txt valid to 7 November 2026, a bug bounty, SOC 2 Type II and ISO 27001 reports for Cloud API, HMAC-signed webhooks and optional mutual TLS (19).

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.

## 3. Maintenance & community, 34 out of 100, up to 5.8 more on the total

Why it scored 34: Closed service. The newest dated API change we found is the OpenAPI document's update on 11 August 2026, 58 days ago, after Graph API v26.0 on 29 July (20). Two dated changes in the 90 days to 8 October, and the WhatsApp changelog's last entry is 12 May 2026, so the three-entry line isn't met (0). A changelog, developer support, a community forum, bug reports and Direct Support with a 24-hour first-response aim for partners. The changelog has one entry in 2026 (8). No current official SDK. The Node.js SDK is archived at 0.0.5-Alpha from April 2023, and the MCP server isn't in the official MCP registry. A Postman collection and the OpenAPI document exist (3). The OpenAPI repository is active but still at v23.0 (3).

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. Agent ergonomics, 67 out of 100, up to 5.4 more on the total

Why it scored 67: Graph API reads take `fields` and `limit`, and one field, `health_status`, summarises whether a number can send. The MCP server isn't what we graded (20). Cursor paging on list edges and filters on template and analytics reads. Inbound messages can't be listed or paged and arrive only by webhook (18). Errors carry `code`, `error_data.details` and `fbtrace_id`, and the reference maps codes to causes and fixes. Some failures arrive only by webhook (18). No idempotency key for sends was found in the docs or the OpenAPI document, and webhook receivers must deduplicate. The MCP send tool asks for confirmation (5). A text send needs four fields. Receiving needs a public HTTPS server, and Meta has no current SDK for Cloud API since archiving its Node.js one (6).

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.

## 5. Reliability, 80 out of 100, up to 4 more on the total

Why it scored 80: Graded on the public Cloud API, with the hosted lines. metastatus.com lists the WhatsApp Business Platform with a Cloud API component, an incident history and availability and latency figures (20). The Cloud API component shows no incident in the 90 days to 8 October 2026. The Marketing Messages API had high disruptions for 70 minutes on 19 July, and Meta Business Agent a 20-minute low disruption on 1 October, so we departed from the full 30 (25). Before the window, Cloud API had high disruptions for about 2 hours 40 minutes on 12 June and medium disruptions on 23 June. Limits are published with numbers for throughput, per recipient and management calls (15). Throttling codes are documented with a 4^X second backoff rule for the per-recipient limit. No Retry-After header on message sends and no idempotency key were found (10). No SLA found. Meta's terms on facebook.com couldn't be read (0). Cloud API is generally available, while the MCP server is in beta (10).

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.

## 6. Schema & documentation, 82 out of 100, up to 2.9 more on the total

Why it scored 82: A public OpenAPI 3.1 document in facebook/openapi with 78 paths, 113 operations and 369 schemas. It is pinned to v23.0 while the Graph API is at v26.0, so 3 off (22). llms.txt indexes and a Markdown copy of every guide and reference page (10). Guides explain when a template is required, when the service window applies and which token type suits which developer. Reference text is thin in places, and the send endpoint's description reads only Send Message (13). Request schemas carry types, required marks and 201 enum lists, with templates and interactive messages as nested objects (12). Request examples on most pages, example webhook payloads and an error code reference. Examples mix v17.0, v23.0 and v25.0 (13). Graph API versions with a dated table, and a WhatsApp changelog whose only 2026 entry is 12 May (12).

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.

## 7. Transparency & trust, 82 out of 100, up to 1.6 more on the total

Made of editorial 70, provenance 94.

Why it scored 82: Editorial half only. Closed service with named terms and policies and an MIT OpenAPI document. The Meta Terms for WhatsApp Business Platform render only with JavaScript and weren't read (15). The docs state Meta is processor for Cloud API, messages are kept at most 30 days and encrypted at rest, and the data processing terms of 22 August 2025 agree on processor duties, breach notice and deletion on termination (25). Graph API versions carry end dates about two years out, the pricing calendar fixes one, three and six months' notice, and deprecated fields are flagged in the docs (20). Processing is in Meta data centres with optional in-country storage. The data processing terms allow Meta companies and third parties as sub-processors, and the list is said to be in the hosting terms, which we couldn't read (10).

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):

- Terms of service: published, but our reader couldn't read it (7 of 10)
- Privacy policy: read, states 4 of the 8 things a reader expects (7 of 10)

## 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.

- unchecked: the Meta Terms for WhatsApp Business Platform and the Cloud API hosting terms on www.facebook.com render only with JavaScript, so the terms, any SLA, the sub-processor list and the wording of the AI Providers clause weren't read
- unchecked: the rate card CSV and PDF files. The docs' Markdown copy drops their links, and the interactive table on whatsappbusiness.com loads rates from an endpoint that needs a page token, which we didn't use
- unchecked: GitHub stars and open issues for facebook/openapi. The GitHub API refused us for its rate limit
- unchecked: the MCP server's tool schemas and annotations, which need a signed-in Meta account to list. Tool names come from the documentation
- unchecked: the official Postman workspace, which renders only with JavaScript
- unchecked: WhatsApp entries in the Graph API v26.0 changelog page, which wasn't opened
- The main pricing page says non-template messages are free, while the non-template pricing page says service messages are charged from 1 October 2026. We recorded the later page's timeline
- provenance.privacy points at the WhatsApp Business Data Processing Terms, which we read. Whether the Cloud API hosting terms are the better governing document couldn't be settled without reading them
- The AI Providers restriction is taken from Meta's pricing page describing the terms. An operator running a general-purpose assistant should read the terms before building
- No idempotency key, no Retry-After header on message sends and no SLA were found in the reviewed documentation
- Not tested with a live account, as we had no Meta developer app
- The lead held up. The docs have moved from /docs/whatsapp/cloud-api to /documentation/business-messaging/whatsapp, and the lead didn't mention the OpenAPI document or the beta MCP server

## Weaknesses

- A person has to create a Meta account, a developer app, a business portfolio and a system user in a browser before any token exists
- No idempotency key for message sends was found in the reviewed documentation or the OpenAPI document, so a retried send can reach the user twice
- Meta's pricing page says the 15 January 2026 terms permit general-purpose AI assistants only where Meta is legally required to allow them
- Outside the 24-hour customer service window only pre-approved templates can be sent, and a new business portfolio is limited to 250 recipients in 24 hours
- Meta archived its Node.js SDK, and the published OpenAPI document is pinned to v23.0 while the Graph API is at v26.0

## 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.

- Check when the user last wrote. Free-form messages are accepted only within 24 hours of the user's last message. After that, send an approved template.
- Record the message id from each send and match it to `messages` webhook statuses. Many failures arrive only by webhook, and no idempotency key was found.
- On error 130429 or 131056 wait and retry. Meta's guidance for the per-recipient limit is 4^X seconds, with X rising by one per failure.
- Verify `X-Hub-Signature-256` with the app secret on every webhook, and deduplicate, because failed deliveries are retried for up to 7 days.
- Request `health_status` on the phone number before a campaign to see whether the app, business, account, number and template can send.

## 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.