Mailchimp Transactional (Mandrill)

# Fix list: Mailchimp Transactional (Mandrill)

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/mandrill, the October 2026 research run, assessed 8 October 2026. Grade C, 58.1 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 Mailchimp Transactional (Mandrill): 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, 53 out of 100, up to 9.4 more on the total

Why it scored 53: Graded on the hosted Transactional API with the hosted lines. status.mailchimp.com lists Mailchimp Transactional as its own component and keeps event history by month (20). In the 90 days to 8 October 2026 it shows three Transactional incidents, all sending delays. They ran about 4 hours 8 minutes on 1 September for some accounts, 4 hours 26 minutes on 15 September and 38 minutes on 17 September. None is described as the API being down, and we count the two long delays together as one major (10). Published numbers cover `/messages/search` at 20 calls a minute, test mode at 10,000 messages a day and the demo at 25 emails an hour. The hourly sending quota is set from reputation with no figure, and no general request limit was found (8). The search endpoint returns 429 and the release notes say to retry 60 seconds after the first of the 20 calls. Mail over the hourly quota is queued in a backlog and not refused. No `Retry-After` header and no idempotency key for sends were found (5). No SLA found, and the additional terms supply the service as-is (0). 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.

## 2. Payments & pricing, 35 out of 100, up to 8.1 more on the total

Why it scored 35: No x402, MPP or L402 (0). Block prices are public without a login, $20 per 25,000 emails falling to $10, and $29.95 a month for a dedicated IP. Full use also needs a Standard or Premium marketing plan, whose price is on a separate page (20). The demo gives 500 emails once on any Mailchimp account, free plan included. We take 5 off because it sends only to the account's own verified domain and we could not confirm that signup asks for no card (15). A person signs up in a browser, enables the add-on, creates a key and sets DNS records, so there is no autonomous route (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.

## 3. Agent ergonomics, 60 out of 100, up to 6.5 more on the total

Why it scored 60: Graded on the API. `/messages/search` takes a `limit` up to 1,000 and send responses are one small object per recipient. No field selection was found. The hosted MCP server has 9 tools and reaches the 99 operations through `list_api`, `describe_api` and `call_api`, which keeps the definitions small (16). Search filters by date, tag, sender and API key, with no offset or cursor, so larger result sets go through a queued export (10). Errors return `status`, `code`, `name` and `message`, and the fundamentals page pairs common ones with a cause and a fix (16). No idempotency key. Test keys let an agent try sends without delivery, and scheduled messages can be listed and cancelled. We could not read the MCP tool annotations without a key (5). A send needs a key, a sender, a recipient and content. Official libraries for Node, Python, PHP and Ruby (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.

## 4. Maintenance & community, 28 out of 100, up to 6.3 more on the total

Why it scored 28: The newest dated release we found is version 1.4.1 of the specification and the Node and Python libraries on 24 March 2026, 198 days before this check (0). No dated Transactional release or changelog entry in the last 90 days. The newest changelog entry is 19 August 2025 (0). A public changelog, a support contact and an API help address in the specification. We saw no public issue replies (6). The Node and Python libraries are both at 1.4.1 and match the current API version, with about 226,000 and 132,000 weekly downloads (15). The generator repository has workflows that validate the specification and build and test each library, and the Node library was updated in February 2024 for an Axios advisory. The test workflow needs a secret key, so we could not see results (7).

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.

## 5. Security & auth, 68 out of 100, up to 5.6 more on the total

Why it scored 68: API keys are created and revoked in the web app, several to an account, and each can be limited to chosen endpoints and to IP addresses. Keys do not expire and no rotation tooling was found. A new key has full account access by default. The key travels in the JSON body or a Bearer header, not the URL (26). Per-endpoint key permissions, IP limits, test keys that send nothing and subaccounts that can be paused. No approval step for sends or deletes (15). The API and the MCP server return inbound mail and stored message content, which is untrusted text. No guidance on prompt injection was found in the MCP guide or the docs (3). The Account Security page lists the IP addresses that used the account and whether each came by API, SMTP or web, and searches can filter by API key (10). A responsible disclosure programme with no rewards, a security.txt with a contact and no Expires field, SOC 2 reports and ISO 27001 per the security page, and regular penetration tests. No public advisory feed found (14).

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.

## 6. Schema & documentation, 79 out of 100, up to 3.4 more on the total

Why it scored 79: An OpenAPI 3.1 file with all 99 operations is public at mailchimp.com/developer/openapi/api-reference-2.json, and a Swagger 2.0 copy sits in the client library repository. The link to it in the Transactional llms.txt returns 404 (25). llms.txt for the developer site and for Transactional, and a Markdown copy of every page at the URL plus `.md` (10). Each operation has a summary and a description averaging about 120 characters. They say what a call does and seldom when not to use it (11). Request bodies carry 36 enums, required fields, date patterns and minimum and maximum values. Headers, metadata and merge variables are free-form objects (11). Reference pages show request and response samples with code in eight languages, and each operation lists typed error bodies. The OpenAPI file itself has no examples (13). The version is in the path, but the docs use 1.0, 1.3 and 1.4 in different places, and the public changelog has nothing newer than 19 August 2025, so the move to 1.4 in March 2026 is not in it (9).

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, 71 out of 100, up to 2.5 more on the total

Made of editorial 64, provenance 77.

Why it scored 71: Closed service with published terms and a named contracting entity. The client libraries are source-available under Mailchimp's own licence agreement, which is not an OSI licence (15). The additional terms and the docs agree that sent message content is kept 30 days, the docs add 90 days for bounce details and 48 hours for the backlog, and a data processing addendum dated 26 September 2024 covers deletion. The terms add that retention is not guaranteed (24). No deprecation policy was found, and the API Use Policy says the APIs can change by a posting or an email. Past notices carried dates, such as the TLS 1.0 and 1.1 retirement announced on 5 June 2023 for 18 July 2023 (10). A sub-processor list with locations is public but was last updated on 29 February 2024, and the security page places Mailchimp's own servers in United States data centres (15).

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):

- Endpoint on the vendor's domain: mandrillapp.com is not on mailchimp.com (0 of 15)
- Terms of service: read, states 5 of the 7 things a reader expects, and has 2 clauses that cost points (4.3 of 10)
- Privacy policy: read, states 8 of the 8 things a reader expects, and has 1 clause that costs points (8 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 price of the Standard and Premium marketing plans that Transactional requires. The marketing pricing page draws its prices by script
- unchecked: whether signing up for a free Mailchimp account to use the demo asks for a payment card
- unchecked: the MCP server's tool schemas and annotations, which need an API key to list
- unchecked: the key settings page at mandrillapp.com/settings, which is behind a login. Per-endpoint and per-IP key limits are taken from the fundamentals page's error table and the MCP guide
- unchecked: the detail pages of the three September 2026 status events, so scope and cause beyond the summaries are not known
- Whether any Transactional API change shipped after 24 March 2026. The changelog's newest entry is 19 August 2025 and the docs pages carry no dates
- Security incidents in the last 12 months were not searched for beyond the vendor's own pages, because web search was not available
- The official MCP registry has no Mailchimp or Mandrill entry under a vendor namespace

## Weaknesses

- Full use needs a Standard or Premium Mailchimp marketing plan in addition to the email blocks
- No idempotency key on `/messages/send`, and no `Retry-After` header found in the docs or the specification
- Three delivery-delay incidents in September 2026, two of them lasting more than four hours
- The newest client library and specification release is 24 March 2026, and the changelog's newest entry is 19 August 2025
- The docs give three API version paths (1.0, 1.3 and 1.4), and the llms.txt link to the OpenAPI file returns 404

## 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.

- Send every call as POST with the key in the JSON body as `key`. The MCP server takes the same key as `Authorization: Bearer`
- Use a test API key while building. Test sends are free, capped at 10,000 a day, and deliver nothing
- Verify the sending domain with DKIM and DMARC first. Mail from an unverified domain is rejected with the reason `unsigned`
- Keep `/messages/search` under 20 calls a minute and set `limit` (default 100, maximum 1,000). There is no offset, so use an activity export for more
- For a restricted key that must also use the MCP server, enable the AI Agents permission group on the key

## 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.