Gmail API

# Fix list: Gmail API

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/gmail-api, the October 2026 research run, assessed 8 October 2026. Grade BB, 77.8 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 Gmail 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, 35 out of 100, up to 8.1 more on the total

Why it scored 35: No x402, MPP or L402 (0). Free within published quotas, with a unit cost for every method, but the price for use above 80,000,000 units a day, planned for later in 2026, isn't published on the quota page or the standardised-model page on 8 October (15). No charge for standard use and no card needed to enable the API (20). A person creates a Cloud project and an OAuth consent screen, a user consents in a browser, and a public app with restricted scopes goes through verification and a security assessment (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, 81 out of 100, up to 3.3 more on the total

Why it scored 81: OAuth 2.0 with 14 scopes and revocable tokens, from `gmail.labels` and `gmail.send` up to the full `https://mail.google.com/` scope. The discovery document lists `access_token` and `oauth_token` as query parameters, and Google's OAuth guide shows the query string as an alternative while calling the header preferable, so the checklist's 10-point deduction applies (30 less 10, 20). `gmail.readonly`, `gmail.metadata` (headers and labels, no bodies), `gmail.labels` and `gmail.send` for least privilege, and permanent delete only under the full scope. Nothing in the API confirms a send or a delete, and every scope that reads mail is restricted (17). The MCP guide warns about indirect prompt injection through email, requires screening of prompts and responses with Model Armor or a documented alternative, and tells operators to review actions. The REST guides say nothing on it (12). Workspace OAuth log events record the API name, method, OAuth client ID and scopes, with the investigation tool limited to higher editions, and no per-call log for personal accounts was found (12). security.txt valid to 2030 with the Google VRP, and the Google Workspace SOC 2 Type II reports, issued quarterly, list Gmail and the Gmail REST API in scope. Apps holding restricted data must pass a CASA assessment every 12 months (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.

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

Why it scored 82: Google publishes a discovery document, not OpenAPI, which is a machine-readable contract all the same, at revision 20260928 with 79 methods on 8 October. The preview MCP server's 23 tools each carry typed input and output schemas (25). No llms.txt, and developers.google.com/llms.txt and /workspace/llms.txt returned 404 on 8 October (0). Method descriptions say what each call does and sometimes what to call in its place, such as `messages.delete` pointing to `messages.trash`, and the MCP tool descriptions name the tool to use first (16). Typed parameters with enums such as `format` and defaults such as `userId=me`, but a sent message is one base64url RFC 2822 string in `raw` and search is one free-text `q` string (11). An error guide with a JSON sample and a fix for each reason, and code samples across the guides (15). v1 in the path and dated release notes, though the last entry for the core API is 1 May 2026 and the growth of the MCP server from 10 tools to 23 isn't recorded there (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.

## 4. Agent ergonomics, 82 out of 100, up to 2.9 more on the total

Why it scored 82: Responses size with `format=minimal` or `metadata`, `metadataHeaders`, `maxResults` and Google's standard `fields` partial responses. List calls return IDs only, so reading mail takes a second call for each message, which batching of up to 100 calls eases. The MCP preview's 23 tool definitions come to about 71 KB (20). `pageToken`, the Gmail search syntax in `q`, `labelIds` and `history.list` for changes since a stored `historyId` (20). Error reasons with a documented fix, fewer of them than the Calendar and Drive guides list (17). Label changes and trash are safe to repeat and the MCP tools carry readOnlyHint, destructiveHint and idempotentHint, but sends have no idempotency key (12). Most parameters are optional and client libraries cover the main languages, but a send needs a hand-built MIME message (13).

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, 90 out of 100, up to 2 more on the total

Why it scored 90: Graded on the REST API with the hosted lines. The Google Workspace Status Dashboard lists Gmail with an incidents feed that goes back to October 2025 (20). In the 90 days to 8 October 2026 the feed has one Gmail entry, a low-severity notice from 18 to 24 September about sign-in on the Gmail Android app, with no entry that touches the service or the API. The last service disruption for Gmail was 56 minutes on 31 May 2026, across several Workspace products. The dashboard tracks the Gmail product, not the API alone (30). 1,200,000 quota units a minute per project, 6,000 a minute per user and 80,000,000 a day per project, with a unit cost for every method (15). 403 and 429 rate errors with a truncated exponential backoff formula, and 429s for sending and bandwidth limits that carry a retry time. `messages.send` has no idempotency key, and the error guide says a 200 response doesn't mean the mail was sent (10). The Workspace SLA covers Gmail at 99.9 per cent but measures downtime in the user web interface and doesn't name the API, read the same way as for Drive (5). The REST API is GA. The MCP server is a developer preview (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. Transparency & trust, 82 out of 100, up to 1.6 more on the total

Made of editorial 69, provenance 94.

Why it scored 82: Closed service under the Google APIs Terms of Service (last modified 9 November 2021), with Apache-2.0 client libraries (15). The Cloud Data Processing Addendum commits Google to delete Workspace customer data within a maximum of 180 days of a deletion, and the Workspace developer policy limits what third-party apps may do with Gmail data, including no transfer for advertising and no training of general models. We found no retention statement for personal accounts' data reached through the API, and didn't read the privacy policy text (20). Dated announcements ahead of changes, such as existing projects keeping their quotas for at least 60 days after 1 May 2026 and at least 90 days' notice before quota charges, but no stated deprecation period, and the API terms let Google discontinue an API (14). The Workspace sub-processor list, last modified 23 July 2026, names third-party sub-processors with the country each processes in, and covers Gmail as a Core Service for Workspace tenants (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):

- Terms of service: read, states 6 of the 7 things a reader expects, and has 1 clause that costs points (7.1 of 10)
- Privacy policy: read, states 7 of the 8 things a reader expects, and has 1 clause that costs points (7.3 of 10)

## 7. Maintenance & community, 85 out of 100, up to 1.3 more on the total

Why it scored 85: @googleapis/gmail 22.0.1 on 23 September 2026, and discovery revision 20260928 (30). Four client releases in the 90 days, 20.0.0 and 21.0.0 on 10 September, 22.0.0 on 14 September and 22.0.1 on 23 September, though the Gmail API release notes have no entry after 24 June 2026 (20). Dated release notes with a feed and public client repositories with commits to 7 October. We didn't sample replies on the issue tracker (12). Current official client libraries, with google-api-python-client 2.201.0 on 30 September (15). Generated clients rebuilt from discovery with a CI workflow in google-api-nodejs-client, whose pass state we didn't check. The repository's own copy of the Gmail discovery file was at revision 20260727 (8).

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.

## 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.

- Price per quota unit above the 80,000,000-a-day threshold, and the date charging starts
- Which scopes the MCP preview's label, trash and spam tools need, given the setup guide configures only `gmail.readonly` and `gmail.compose`
- Why the live MCP server returns 23 tools while the reference page lists 10 and the setup guide 9
- Whether gmail.googleapis.com still accepts an access token in the query string. We read the discovery document and the OAuth guide and made no authenticated call. The same parameters appear for other Google APIs, and the Drive and Calendar dossiers didn't take this deduction
- Whether failures of the Gmail API alone count as downtime under the Workspace SLA, which measures the user web interface
- What a CASA security assessment costs
- unchecked: PyPI weekly downloads for google-api-python-client (pypistats.org answered 429 on 8 October)
- unchecked: the text of Google's privacy policy (the page answered 200 but we didn't read it) and a retention statement for personal accounts' data reached through the API
- unchecked: replies on Google's public issue tracker for the Gmail API, and the pass state of CI in google-api-nodejs-client
- unchecked: user caps and token lifetimes for unverified apps in testing, which sit in the OAuth verification help centre pages we didn't load
- unchecked: the daily sending limit for personal gmail.com accounts, which the Workspace limits page doesn't give

## Weaknesses

- Eight restricted scopes, including `gmail.readonly` and `gmail.metadata`, need restricted-scope verification for a public app
- An app that stores or transmits restricted data on servers needs a security assessment by a Google-approved assessor every 12 months
- No idempotency key on `messages.send`, and the error guide says a 200 response doesn't confirm the mail was sent
- MCP server limited to the Developer Preview Programme, with no send tool
- Price for use above the daily quota not yet published

## 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.

- Ask for the narrowest scope. `gmail.labels` is non-sensitive and `gmail.send` is sensitive, while every scope that reads mail is restricted
- List with `messages.list` and `q` in Gmail search syntax, then fetch each ID with `format=metadata` or `full`. List calls return IDs only
- Send by posting an RFC 2822 message, base64url encoded, in `raw`. Set `threadId` and matching reply headers to stay in a thread
- Store the `historyId` and call `history.list`. On a 404, run a full sync. Call `watch` again at least every 7 days
- Back off exponentially on 403 and 429 rate errors, and keep batches to 50 requests or fewer

## 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.