Poke

# Fix list: Poke

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/poke, the October 2026 research run, assessed 12 October 2026. Grade F, 37.6 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 Poke: 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. Security & auth, 24 out of 100, up to 13.3 more on the total

Why it scored 24: API keys are named, created in Kitchen, shown once and revoked individually, with no scopes or expiry (20). We took 5 off, half the checklist's deduction, because the docs give a prefilled link that carries a third-party MCP server's key as `apiKey` in the query string. The secret is not Poke's own key, so not the full 10 (15 of 30). A message sent with a key is processed with full access to email, calendar and integrations. No read-only mode or approval setting is documented. Release notes mention email draft approval and attachment preview before sending, without describing a control (4 of 20). Poke reads mail, web pages and MCP tool output, and no prompt-injection guidance was found (0 of 15). API messages appear in the conversation, and MCP servers receive `X-Poke-User-Id`. No audit log is documented (3 of 15). security.txt answers 404, and no disclosure policy, bounty or certification was found on poke.com. The privacy notice commits to Google's Limited Use requirements (2 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.

## 2. Agent ergonomics, 30 out of 100, up to 11.4 more on the total

Why it scored 30: Graded on the Poke API and the `poke` npm package. A call returns a two-field acknowledgement and the assistant's answer goes to the owner's message thread, so a calling agent cannot read the result (10 of 25). No endpoint reads conversations, memories or integrations, so there is nothing to page or filter (3 of 20). Errors documented are 401 and a `success` boolean (6 of 20). No idempotency key or retry guidance, and a repeated POST is a second instruction (2 of 20). One endpoint and one conventional field. One official SDK, for Node.js, which also creates webhook triggers (9 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. Reliability, 51 out of 100, up to 9.8 more on the total

Why it scored 51: Read as a hosted service. status.poke.com is an incident.io page with components for App, Email, Integrations, iMessage, WhatsApp, SMS and API, and a history view (20). Two incidents from July to 12 October 2026. On 22 July iMessage had a 42-minute partial outage attributed to an upstream provider. An issue with Poke Sites, marked degraded performance on App, stayed open from 15 to 29 September and closed with the note "fixed it". No core outage of an hour or more, but one incident open for two weeks (18 of 30). The API page has a rate limits heading with no numbers (0). It advises batching and delays, with no `Retry-After`, backoff figures or idempotency key (3 of 15). No SLA found (0). The release notes of 19 March 2026 call Poke generally available (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.

## 4. Schema & documentation, 42 out of 100, up to 9.4 more on the total

Why it scored 42: Graded on the Poke API, the one documented interface an outside program can call. No OpenAPI file or other machine-readable contract was found (0). poke.com/llms.txt lists seven docs pages (10). The API page states the purpose, four use cases and what happens after a message arrives (12 of 20). The body is any JSON object, forwarded whole to the assistant, with `message` shown by convention and no types or constraints (2 of 15). Examples in curl, Python and JavaScript, a response sample, and one documented error, 401 (9 of 15). The path carries `v1`. The page marks the older `/api/v1/inbound-sms/webhook` deprecated with no date. Release notes are public and have covered major updates only since March 2026 (9 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.

## 5. Payments & pricing, 30 out of 100, up to 8.8 more on the total

Why it scored 30: No x402, MPP or L402 in the docs, llms.txt or pricing page (0). Plan prices are public, $0, $19 and $199 a month, and Ultra credits are $1 each. Allowances are not given in numbers and nothing is priced per call (10). The Free plan needs no card (20). A person signs up with a phone number or email and creates the key 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.

## 6. Maintenance & community, 31 out of 100, up to 6 more on the total

Why it scored 31: The newest product entry in the release notes is Poke Human on 25 June 2026, 109 days before the check (10 of 30). The entry of 23 July 2026 announces the move to Cognition and changes nothing in the product. Counting it would give 20. One dated entry in the last 90 days (0). A public changelog, a support address, a Discord server and a status page whose incidents close with one-line notes (8 of 15). The `poke` npm package, a CLI and Node.js SDK, reached 4.0.461 on 23 August 2026 through GitHub Actions trusted publishing, the only official SDK (8 of 15). The package declares no dependencies and links no repository (5 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.

## 7. Transparency & trust, 58 out of 100, up to 3.7 more on the total

Made of editorial 34, provenance 81.

Why it scored 58: Closed service with published Terms of Service, last modified 17 March 2026, naming The Interaction Company of California d/b/a Poke (15 of 30). The privacy notice of 11 September 2026 gives retention criteria and no periods. It says customer content, including email contents, may train AI models unless the user selects Maximum Privacy, and the terms allow training third-party models. Google account data is limited to user-facing functions under Google's Limited Use rules. The FAQ says account deletion takes up to 24 hours, with backups later. No DPA found (12 of 30). One endpoint is marked deprecated with no date, and no deprecation policy was found (4 of 20). Stripe and Google Analytics are named. Model providers, hosting and other sub-processors are not (3 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):

- Terms of service: read, states 6 of the 7 things a reader expects, and has 4 clauses that cost points (1.1 of 10)
- security.txt: not found (0 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.

- Kind is `platform`. Poke is a hosted assistant used through messaging apps, and its API is a single inbound message endpoint that returns no result, so `http-api` would overstate it. Graded as a hosted service.
- The lead gave no docs link and described no public interface. Poke has docs at poke.com/docs, an inbound API, an npm CLI and SDK, and RCS as a fourth channel.
- Maintenance scores recency on the Poke Human entry of 25 June 2026. Counting the Cognition notice of 23 July 2026 would add 10.
- The Security deduction of 5 for the `apiKey` query parameter on the prefilled integration link is a judgement call. The checklist's 10 is written for a product's own credential.
- The terms (section 2.2) forbid automated access for scraping and use for benchmarking or competitive analysis. Recorded as a fact with no deduction. It matters before any probe is run.
- The terms and privacy notice do not mention Cognition. Whether Cognition's policies or models now apply to Poke data was not established.
- unchecked: which actions Poke takes without approval and any approval setting, which sit in the signed-in app and were not read.
- unchecked: the Maximum Privacy setting, its default and what else it changes. The privacy notice links the FAQ for privacy protections, and the FAQ page read has no privacy section.
- unchecked: model providers, sub-processors, hosting region and credential storage. None is named on the pages read.
- unchecked: OAuth scopes requested from Google and Microsoft. The privacy notice says the narrowest needed, without listing them.
- unchecked: poke.com/health (consumer health privacy notice), poke.com/docs/merging-accounts, the recipe pages and poke.com/explore, left unread to keep to the page budget.
- unchecked: a source repository for the `poke` npm package. The registry record links none, and no GitHub organisation was confirmed.
- unchecked: npm weekly downloads and whether the Free plan includes API keys.
- unchecked: security certifications. The site loads a cookie consent script from a compliance vendor, which is not evidence of a certification.
- poke.com/robots.txt and status.poke.com/robots.txt answered 404, read as no rules. registry.npmjs.org/robots.txt answered with a package record, not a robots file, read as no rules.

## Weaknesses

- No documented setting for which actions need approval before Poke sends, deletes or spends. Release notes mention email draft approval only in passing
- Customer content, including email contents, may train AI models unless the user selects Maximum Privacy, per the privacy notice of 11 September 2026
- Model providers, sub-processors, hosting locations and where account tokens are stored are not named in the pages read
- An API key message is processed with full access to email, calendar and integrations. Keys have no scopes
- No security.txt, disclosure policy, bug bounty or certification was found on poke.com
- The terms forbid automated access for scraping and use for benchmarking or competitive analysis. This matters before any probe is run

## 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.

- POST JSON to `https://poke.com/api/v1/inbound/api-message` with a Bearer key from Kitchen. Legacy `pk_` keys work only on the deprecated `/api/v1/inbound-sms/webhook`
- Expect only `{"success": true}` back. Poke's answer goes to the owner's message thread, not to the caller
- Do not retry a POST blindly. No idempotency key is documented, and a repeated message may make Poke act twice
- Treat an API key as full access to the owner's mail, calendar and integrations. Create one key per automation and revoke it when done
- Avoid the prefilled integration link with `apiKey` in the query string for real secrets. Enter the key in the form or use `npx poke@latest mcp add -k`

## 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.