{
  "fixes": {
    "slug": "pleo",
    "name": "Pleo API + MCP",
    "listing": "https://www.anchorterminal.com/tools/pleo",
    "markdown": "# Fix list: Pleo API + MCP\n\nFrom Anchor Terminal's listing at https://www.anchorterminal.com/tools/pleo, the October 2026 research run, assessed 8 October 2026. Grade B, 62.9 out of 100.\n\nThis 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.\n\nFor a coding agent working on Pleo API + MCP: 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.\n\n## 1. Payments \u0026 pricing, 15 out of 100, up to 10.6 more on the total\n\nWhy it scored 15: Read with the hosted rubric. No x402, MPP or L402 (0). Per-user plan prices are public (£8, £14 and £18 a month as shown to a UK visitor), with no per-call price for the API or MCP (10). The signup form states 21 days free on Start and Build, but the pricing page lists MCP on Optimise only, which is sold through a demo, and API keys need enabling by Pleo support. Whether the trial needs a card isn't stated. Partial credit (5 of 20). A person signs in with OAuth in a browser or creates a key in the web app, and Pleo issues partner OAuth clients after review (0).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):\n\nThe published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).\n\n- 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.\n- 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.\n- 20, a free tier or trial that doesn't need a card.\n- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).\n\nPayment 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.\n\nOpen-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.\n\n## 2. Agent ergonomics, 55 out of 100, up to 7.3 more on the total\n\nWhy it scored 55: The MCP tool count and definitions aren't published, so context cost couldn't be measured. On the API, `limit` sizes a page and aggregated endpoints return tag groups and category groups in one call, with no field selection (13 of 25). Cursor and offset pagination with sorting keys, and search endpoints with filters for accounting entries, receipts, tags, vendors and employees. Two pagination styles are in use (17 of 20). Errors carry a `type` code and a message, and every response has a request-Id header. The general error page lists seven statuses with generic fixes and leaves out 409, 422 and 429 (13 of 20). No Idempotency-Key. The docs ask integrators to make their own processing idempotent, and export jobs can be resumed. MCP readOnlyHint and destructiveHint annotations couldn't be read (7 of 20). List calls need few parameters. No official SDK was found, only a Postman collection and two community libraries named for the Legacy API (5 of 15).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):\n\n- 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).\n- 20, pagination, filtering and output-size controls.\n- 20, actionable, documented error responses, codes and messages an agent can recover from.\n- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.\n- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.\n\nModels 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.\n\n## 3. Reliability, 71 out of 100, up to 5.8 more on the total\n\nWhy it scored 71: Read with the hosted lines and scored on the MCP server and the External API together. status.pleo.io on incident.io has 25 components with incident history, among them Pleo API (20). From 10 July to 8 October 2026 the page lists two incidents, both marked minor and neither on the Pleo API component. Card transactions failed intermittently for about 2 hours 30 minutes on 15 July during a Mastercard authentication fault, and instant top-up in Sweden was impaired from 10 to 15 September (20). One limit is published, 600 requests a minute per credential across all endpoints. No limit is documented for the MCP server (15). The docs tell clients to wait for Retry-After when present, otherwise back off from one second, and say which steps are unsafe to repeat after a 429. There is no Idempotency-Key header, and the OpenAPI specs don't declare 429 (11 of 15). No SLA found in the API terms or the UK Master Service Agreement (0). The API terms describe the API as a beta version, and the specs include /v0 and /v1-beta paths. The MCP server carries no beta label (5 of 10).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):\n\nHosted APIs, MCP servers, models and platforms.\n\n- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).\n- 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.\n- 15, rate limits documented with numbers.\n- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.\n- 10, an SLA published for any paid tier.\n- 10, the surface agents use is generally available, not beta or preview.\n\nLocal packages, SDKs, frameworks and stdio MCP servers.\n\n- 20, installs from an official package with supported runtimes stated.\n- 25, a public CI and test suite, passing on the default branch.\n- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).\n- 15, semver discipline and breaking changes called out in a changelog.\n- 15, version 1.0 or later, or declared stable.\n\nProtocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.\n\n## 4. Security \u0026 auth, 71 out of 100, up to 5.1 more on the total\n\nWhy it scored 71: The MCP server uses OAuth authorisation code with PKCE (S256), refresh tokens and dynamic client registration, with a single scope and the connecting user's Pleo role as the real boundary. The API uses OAuth 2.0 with resource scopes and rotating refresh tokens, or API keys with chosen scopes and an expiry, sent in the Authorization header (28 of 30). Read and write are separate API scopes. MCP access is off until an admin enables it per entity, can't be limited to named users yet, and has no read-only mode. Payments, card changes and limit changes are blocked through the MCP by design. Confirmation before a write is left to the AI client (15 of 20). Merchant names, notes and receipts are untrusted text. The docs say to review significant output and recommend approval for write tools, with no guidance on injected content (5 of 15). Changes made through the MCP appear in Pleo's activity and audit logs, and users can list and revoke MCP connections. No call log for API keys was found (10 of 15). A disclosure policy with safe harbour, a HackerOne bug bounty, PCI-DSS and Google's CASA per the trust page. No SOC 2 or ISO 27001 is claimed, reports are shared under NDA, and security.txt returns 403 (13 of 20).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-security):\n\n- 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.\n- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.\n- 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.\n- 0 to 15, audit logs or per-call visibility for the operator.\n- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.\n\nModels 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.\n\n## 5. Maintenance \u0026 community, 64 out of 100, up to 3.2 more on the total\n\nWhy it scored 64: The newest changelog entry is 2 October 2026, a new Teams API (30). Sixteen dated entries between 15 July and 2 October 2026, among them the MCP documentation on 17 August and Export API v3 additions (20). Closed service with a dated changelog, api@pleo.io for partners and Pleo support for customers. The docs say custom implementation support is generally not given, and response times couldn't be observed (9 of 15). No official SDK was found, the MCP server isn't in the official MCP registry, and the docs say Pleo isn't in Claude's connector directory yet (0). No packages to assess. The published specs carry current version numbers and match the changelog (5 of 10).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):\n\n- 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.\n- 20, at least three releases or dated changelog entries in the last 90 days.\n- 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.\n- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).\n- 10, package health, current dependencies and CI.\n\nModels are read for deprecation notice periods and model churn rather than release counts.\n\n## 6. Schema \u0026 documentation, 82 out of 100, up to 2.9 more on the total\n\nWhy it scored 82: Eleven current OpenAPI 3.0.1 specs with 162 operations, linked from llms.txt, plus a spec for the deprecated Legacy API (25). llms.txt, llms-full.txt and a Markdown copy of every docs page at the same URL with .md (10). 107 of 162 operations carry a description, and long guides explain when each export and sync step applies. The MCP server has no public tool reference, only prose on what it can do, so its tool descriptions couldn't be read without a customer sign-in. Three specs carry internal titles (Gjoll, Oberon, Triton) (11 of 20). The specs hold 86 enums and required lists, and search endpoints take typed filter bodies. MCP input schemas are unread (11 of 15). 463 examples across the specs and typed error examples such as MISSING_CONTRA_ACCOUNTS and INVALID_TARGET_SYSTEM. 77 operations declare only a default error response and none declares 429 (11 of 15). Path versions, with Export API v1 to v3 documented side by side, and a public changelog of 56 dated entries back to November 2024. The current API has no written versioning policy (14 of 15).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):\n\nAPIs and MCP servers.\n\n- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).\n- 10, llms.txt or Markdown docs served for agents.\n- 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.\n- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.\n- 0 to 15, examples and documented error responses.\n- 15, versioning and a public changelog.\n\nModels 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.\n\n## 7. Transparency \u0026 trust, 75 out of 100, up to 2.2 more on the total\n\nMade of editorial 63, provenance 87.\n\nWhy it scored 75: Closed service with public API terms naming Pleo Technologies A/S, a Master Service Agreement and separate AI Access Terms for the MCP. The API terms carry no date and let Pleo discontinue the API at any time (14 of 30). A privacy notice dated June 2026, a DPA of 14 October 2025 and a trust page placing customer data in AWS Ireland agree with each other. Pleo's MCP page says it doesn't use customer data to train AI models. Retention for customer data is stated only as long as there is a valid purpose or legal requirement, with no periods (20 of 30). The Legacy API is deprecated with migration guides, but no end-of-life date is published, and replacements listed for Q3 2026 (a SCIM API and wallet balance) weren't in the docs index on 8 October. Changes to the API terms take effect no sooner than 30 days after posting (9 of 20). A public sub-processor list with each provider's role and location, and 30 days' notice of new sub-processors in the DPA (20).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):\n\n- 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.\n- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).\n- 0 to 20, a deprecation policy or notices with dates.\n- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).\n\nThe other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.\n\nProvenance checks not met in full (half of this category, computed from checked facts):\n\n- Terms of service: read, states 4 of the 7 things a reader expects (7.4 of 10)\n- security.txt: not found (0 of 10)\n\n## What we couldn't check\n\nWhat 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.\n\n- unchecked: MCP tool names, input schemas, annotations and count. tools/list needs a signed-in Pleo customer and we didn't register a client\n- unchecked: whether the 21-day trial needs a payment card, and whether a trial company can use the MCP server or API keys\n- unchecked: prices in USD. The pricing page showed GBP to our UK fetch, so unitPrices is empty\n- Whether Standalone API Keys are tied to a plan. The pricing page lists MCP by plan but doesn't mention the API\n- Whether the SCIM API and the wallet balance replacement, both listed for Q3 2026, have shipped\n- Whether a rate limit applies to the MCP server, and what its errors look like\n- unchecked: the HackerOne programme's scope and rewards. The trust page states the programme exists and we didn't read the HackerOne page\n- Whether an SLA exists in order forms for larger customers\n\n## Weaknesses\n\n- The pricing page lists MCP on the Optimise plan only (£18 per user per month, three users minimum, sold through a demo)\n- MCP tool names, schemas and count aren't published, so they can't be read without a customer sign-in\n- Standalone API keys aren't self-service. Pleo support or a Customer Success Manager enables them, and partner OAuth clients go through a reviewed programme\n- No Idempotency-Key header, no official SDK and no entry in the official MCP registry\n- The API terms call the API a beta version that Pleo may discontinue at any time, and no SLA was found\n- No end-of-life date is published for the deprecated Legacy API, and its Q3 2026 replacements for employee writes and wallet balance aren't in the docs\n\n## What costs an agent a turn today\n\nThe notes we give agents before they call it. Each one is a workaround an agent shouldn't need.\n\n- Ask a company admin to enable Pleo MCP access under Settings, General, Pleo AI for each entity before connecting. It is off by default\n- Name the entity in every request when working outside the default one. Each MCP request targets one entity and the choice doesn't persist\n- Set the AI client to require approval for Pleo write tools. Pleo leaves confirmation to the client and doesn't enforce it server-side\n- Send API keys as the Basic auth username with an empty password to external.pleo.io. Legacy tokens for openapi.pleo.io don't work there\n- Budget every endpoint against one bucket of 600 requests a minute per credential, and on 429 wait for Retry-After or back off from one second\n- Swap mcp.staging.pleo.io for mcp.pleo.io in the Claude Code command when moving from staging to production. Each needs its own OAuth sign-in\n\n## When it's done\n\nSend 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.\n",
    "grade": "B",
    "score": 62.9,
    "assessed": "2026-10-08",
    "run": "October 2026 research run",
    "categories": [
      {
        "key": "payments",
        "name": "Payments \u0026 pricing",
        "score": 15,
        "maxGain": 10.6,
        "reason": "Read with the hosted rubric. No x402, MPP or L402 (0). Per-user plan prices are public (£8, £14 and £18 a month as shown to a UK visitor), with no per-call price for the API or MCP (10). The signup form states 21 days free on Start and Build, but the pricing page lists MCP on Optimise only, which is sold through a demo, and API keys need enabling by Pleo support. Whether the trial needs a card isn't stated. Partial credit (5 of 20). A person signs in with OAuth in a browser or creates a key in the web app, and Pleo issues partner OAuth clients after review (0).",
        "checklist": [
          "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.\n- 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.\n- 20, a free tier or trial that doesn't need a card.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-payments"
      },
      {
        "key": "ergonomics",
        "name": "Agent ergonomics",
        "score": 55,
        "maxGain": 7.3,
        "reason": "The MCP tool count and definitions aren't published, so context cost couldn't be measured. On the API, `limit` sizes a page and aggregated endpoints return tag groups and category groups in one call, with no field selection (13 of 25). Cursor and offset pagination with sorting keys, and search endpoints with filters for accounting entries, receipts, tags, vendors and employees. Two pagination styles are in use (17 of 20). Errors carry a `type` code and a message, and every response has a request-Id header. The general error page lists seven statuses with generic fixes and leaves out 409, 422 and 429 (13 of 20). No Idempotency-Key. The docs ask integrators to make their own processing idempotent, and export jobs can be resumed. MCP readOnlyHint and destructiveHint annotations couldn't be read (7 of 20). List calls need few parameters. No official SDK was found, only a Postman collection and two community libraries named for the Legacy API (5 of 15).",
        "checklist": [
          "- 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).\n- 20, pagination, filtering and output-size controls.\n- 20, actionable, documented error responses, codes and messages an agent can recover from.\n- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-ergonomics"
      },
      {
        "key": "reliability",
        "name": "Reliability",
        "score": 71,
        "maxGain": 5.8,
        "reason": "Read with the hosted lines and scored on the MCP server and the External API together. status.pleo.io on incident.io has 25 components with incident history, among them Pleo API (20). From 10 July to 8 October 2026 the page lists two incidents, both marked minor and neither on the Pleo API component. Card transactions failed intermittently for about 2 hours 30 minutes on 15 July during a Mastercard authentication fault, and instant top-up in Sweden was impaired from 10 to 15 September (20). One limit is published, 600 requests a minute per credential across all endpoints. No limit is documented for the MCP server (15). The docs tell clients to wait for Retry-After when present, otherwise back off from one second, and say which steps are unsafe to repeat after a 429. There is no Idempotency-Key header, and the OpenAPI specs don't declare 429 (11 of 15). No SLA found in the API terms or the UK Master Service Agreement (0). The API terms describe the API as a beta version, and the specs include /v0 and /v1-beta paths. The MCP server carries no beta label (5 of 10).",
        "checklist": [
          "Hosted APIs, MCP servers, models and platforms.",
          "- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).\n- 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.\n- 15, rate limits documented with numbers.\n- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.\n- 10, an SLA published for any paid tier.\n- 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.\n- 25, a public CI and test suite, passing on the default branch.\n- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).\n- 15, semver discipline and breaking changes called out in a changelog.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-reliability"
      },
      {
        "key": "security",
        "name": "Security \u0026 auth",
        "score": 71,
        "maxGain": 5.1,
        "reason": "The MCP server uses OAuth authorisation code with PKCE (S256), refresh tokens and dynamic client registration, with a single scope and the connecting user's Pleo role as the real boundary. The API uses OAuth 2.0 with resource scopes and rotating refresh tokens, or API keys with chosen scopes and an expiry, sent in the Authorization header (28 of 30). Read and write are separate API scopes. MCP access is off until an admin enables it per entity, can't be limited to named users yet, and has no read-only mode. Payments, card changes and limit changes are blocked through the MCP by design. Confirmation before a write is left to the AI client (15 of 20). Merchant names, notes and receipts are untrusted text. The docs say to review significant output and recommend approval for write tools, with no guidance on injected content (5 of 15). Changes made through the MCP appear in Pleo's activity and audit logs, and users can list and revoke MCP connections. No call log for API keys was found (10 of 15). A disclosure policy with safe harbour, a HackerOne bug bounty, PCI-DSS and Google's CASA per the trust page. No SOC 2 or ISO 27001 is claimed, reports are shared under NDA, and security.txt returns 403 (13 of 20).",
        "checklist": [
          "- 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.\n- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.\n- 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.\n- 0 to 15, audit logs or per-call visibility for the operator.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-security"
      },
      {
        "key": "maintenance",
        "name": "Maintenance \u0026 community",
        "score": 64,
        "maxGain": 3.2,
        "reason": "The newest changelog entry is 2 October 2026, a new Teams API (30). Sixteen dated entries between 15 July and 2 October 2026, among them the MCP documentation on 17 August and Export API v3 additions (20). Closed service with a dated changelog, api@pleo.io for partners and Pleo support for customers. The docs say custom implementation support is generally not given, and response times couldn't be observed (9 of 15). No official SDK was found, the MCP server isn't in the official MCP registry, and the docs say Pleo isn't in Claude's connector directory yet (0). No packages to assess. The published specs carry current version numbers and match the changelog (5 of 10).",
        "checklist": [
          "- 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.\n- 20, at least three releases or dated changelog entries in the last 90 days.\n- 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.\n- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).\n- 10, package health, current dependencies and CI.",
          "Models are read for deprecation notice periods and model churn rather than release counts."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-maintenance"
      },
      {
        "key": "schema",
        "name": "Schema \u0026 documentation",
        "score": 82,
        "maxGain": 2.9,
        "reason": "Eleven current OpenAPI 3.0.1 specs with 162 operations, linked from llms.txt, plus a spec for the deprecated Legacy API (25). llms.txt, llms-full.txt and a Markdown copy of every docs page at the same URL with .md (10). 107 of 162 operations carry a description, and long guides explain when each export and sync step applies. The MCP server has no public tool reference, only prose on what it can do, so its tool descriptions couldn't be read without a customer sign-in. Three specs carry internal titles (Gjoll, Oberon, Triton) (11 of 20). The specs hold 86 enums and required lists, and search endpoints take typed filter bodies. MCP input schemas are unread (11 of 15). 463 examples across the specs and typed error examples such as MISSING_CONTRA_ACCOUNTS and INVALID_TARGET_SYSTEM. 77 operations declare only a default error response and none declares 429 (11 of 15). Path versions, with Export API v1 to v3 documented side by side, and a public changelog of 56 dated entries back to November 2024. The current API has no written versioning policy (14 of 15).",
        "checklist": [
          "APIs and MCP servers.",
          "- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).\n- 10, llms.txt or Markdown docs served for agents.\n- 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.\n- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.\n- 0 to 15, examples and documented error responses.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-schema"
      },
      {
        "key": "transparency",
        "name": "Transparency \u0026 trust",
        "score": 75,
        "maxGain": 2.2,
        "reason": "Closed service with public API terms naming Pleo Technologies A/S, a Master Service Agreement and separate AI Access Terms for the MCP. The API terms carry no date and let Pleo discontinue the API at any time (14 of 30). A privacy notice dated June 2026, a DPA of 14 October 2025 and a trust page placing customer data in AWS Ireland agree with each other. Pleo's MCP page says it doesn't use customer data to train AI models. Retention for customer data is stated only as long as there is a valid purpose or legal requirement, with no periods (20 of 30). The Legacy API is deprecated with migration guides, but no end-of-life date is published, and replacements listed for Q3 2026 (a SCIM API and wallet balance) weren't in the docs index on 8 October. Changes to the API terms take effect no sooner than 30 days after posting (9 of 20). A public sub-processor list with each provider's role and location, and 30 days' notice of new sub-processors in the DPA (20).",
        "blend": "editorial 63, provenance 87",
        "checklist": [
          "- 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.\n- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).\n- 0 to 20, a deprecation policy or notices with dates.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-transparency"
      }
    ],
    "provenance": [
      {
        "label": "Terms of service",
        "value": "read, states 4 of the 7 things a reader expects",
        "points": 7.4,
        "max": 10
      },
      {
        "label": "security.txt",
        "value": "not found",
        "points": 0,
        "max": 10
      }
    ],
    "unchecked": [
      "unchecked: MCP tool names, input schemas, annotations and count. tools/list needs a signed-in Pleo customer and we didn't register a client",
      "unchecked: whether the 21-day trial needs a payment card, and whether a trial company can use the MCP server or API keys",
      "unchecked: prices in USD. The pricing page showed GBP to our UK fetch, so unitPrices is empty",
      "Whether Standalone API Keys are tied to a plan. The pricing page lists MCP by plan but doesn't mention the API",
      "Whether the SCIM API and the wallet balance replacement, both listed for Q3 2026, have shipped",
      "Whether a rate limit applies to the MCP server, and what its errors look like",
      "unchecked: the HackerOne programme's scope and rewards. The trust page states the programme exists and we didn't read the HackerOne page",
      "Whether an SLA exists in order forms for larger customers"
    ],
    "weaknesses": [
      "The pricing page lists MCP on the Optimise plan only (£18 per user per month, three users minimum, sold through a demo)",
      "MCP tool names, schemas and count aren't published, so they can't be read without a customer sign-in",
      "Standalone API keys aren't self-service. Pleo support or a Customer Success Manager enables them, and partner OAuth clients go through a reviewed programme",
      "No Idempotency-Key header, no official SDK and no entry in the official MCP registry",
      "The API terms call the API a beta version that Pleo may discontinue at any time, and no SLA was found",
      "No end-of-life date is published for the deprecated Legacy API, and its Q3 2026 replacements for employee writes and wallet balance aren't in the docs"
    ],
    "agentNotes": [
      "Ask a company admin to enable Pleo MCP access under Settings, General, Pleo AI for each entity before connecting. It is off by default",
      "Name the entity in every request when working outside the default one. Each MCP request targets one entity and the choice doesn't persist",
      "Set the AI client to require approval for Pleo write tools. Pleo leaves confirmation to the client and doesn't enforce it server-side",
      "Send API keys as the Basic auth username with an empty password to external.pleo.io. Legacy tokens for openapi.pleo.io don't work there",
      "Budget every endpoint against one bucket of 600 requests a minute per credential, and on 429 wait for Retry-After or back off from one second",
      "Swap mcp.staging.pleo.io for mcp.pleo.io in the Claude Code command when moving from staging to production. Each needs its own OAuth sign-in"
    ],
    "recheck": "https://www.anchorterminal.com/builders/#disputes"
  },
  "meta": {
    "attribution": "Anchor Terminal (https://www.anchorterminal.com)",
    "docs": "https://www.anchorterminal.com/docs/",
    "generatedAt": "2026-10-08",
    "license": "CC-BY-4.0",
    "method": "https://www.anchorterminal.com/benchmark/",
    "methodology": "0.4",
    "openapi": "https://www.anchorterminal.com/openapi.json",
    "preview": false,
    "run": "2026-10-01",
    "runLabel": "October 2026 research run"
  }
}