Belvo

# Fix list: Belvo

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/belvo, the October 2026 research run, assessed 8 October 2026. Grade B, 63.5 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 Belvo: 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, 20 out of 100, up to 10 more on the total

Why it scored 20: No x402, MPP or L402 (0). The Launch plan is published at 1,000 USD a month and Growth is custom. No per-call or per-link price is public (10 of 20). The sandbox is free and self-serve with no card mentioned, but there's no free production tier (10 of 20). A person signs up in a browser, and production needs a sales meeting and a certification call (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, 57 out of 100, up to 7.5 more on the total

Why it scored 57: One `secretId` and `secretPassword` pair per environment over HTTP Basic, resettable in the dashboard, with no scopes found (18 of 30). Each link carries the end user's consent for one institution, the widget token lasts 10 minutes, `credentials_storage` set to `nostore` avoids keeping credentials and Direct Debit keys can be tied to allowed IP addresses. No read-only key or approval step for payments was found (11 of 20). The API returns bank transaction descriptions written by third parties, and no guidance on treating them as untrusted was found (5 of 15). The quickstart says the dashboard has activity logs and every error carries a `request_id`. We couldn't see the dashboard (8 of 15). ISO/IEC 27001:2022 and PCI DSS Level 1 are stated on the security page with third-party penetration tests, and a PGP-signed security.txt points to a disclosure policy on Federacy, but it expired on 22 March 2025. The data webhooks carry a fixed optional authorisation header and no signature (15 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.

## 3. Maintenance & community, 30 out of 100, up to 6.1 more on the total

Why it scored 30: Closed service. The newest dated product change we found is the launch of the MCP servers, announced on 10 August 2026, 59 days before the check. The Developer MCP reports server version 2026-10-08 and the specification is at 1.223.0, neither with dated notes (20 of 30). No changelog, so no count of dated entries in 90 days (0 of 20). Support by email at support@belvo.com on the lower plans and a feedback widget in the docs. No public channel where answers can be seen (5 of 15). The belvo npm package (0.28.0, 21 June 2023), belvo-python (0.39.1, 21 August 2023) and the belvo gem (1.7.0, 21 June 2023) are stale and their repositories return 404. The Developer MCP is live (3 of 15). Package health couldn't be established without the repositories (2 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.

## 4. Schema & documentation, 80 out of 100, up to 3.3 more on the total

Why it scored 80: Public OpenAPI 3.0.2 file, version 1.223.0, with 146 operations on 83 paths, and a second specification for Direct Debit Mexico (25). llms.txt with about 300 links to Markdown twins of the docs in three languages (10). Operations carry usage sections and notes on limits and asynchronous modes, though the choice between POST retrieval and GET list is explained in the guides (16 of 20). Query parameters are typed and filters are enumerated one by one. We didn't audit request bodies in full (12 of 15). 48 error articles give the JSON body, cause and solution for each code, and the reference documents 13 status codes (13 of 15). The specification has a version number but no changelog was found, and developers.belvo.com/changelog returns 404 (4 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. Agent ergonomics, 80 out of 100, up to 3.3 more on the total

Why it scored 80: List endpoints accept `fields` and `omit` to size responses and `page_size` from the default 100 up to 1,000 (22 of 25). Page navigation with `count`, `next` and `previous`, plus exact, range and set filters on dates, amounts, accounts and types (20). Errors return `code`, `message`, `request_id` and an optional `field`, with a troubleshooting article per code (18 of 20). `Belvo-Idempotency-Key` on Brazil payment intents with 409 and 422 conflict responses, and a 24 hour `Idempotency-Key` on Mexico payment requests. Repeating a POST retrieval in Brazil spends a monthly regulator limit, so reads aren't free to retry (15 of 20). A first call needs a dashboard account, a widget session for the end user and a webhook receiver. The Python, Node and Ruby packages were last published in 2023, their GitHub repositories return 404 and the docs no longer list them (5 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.

## 6. Reliability, 84 out of 100, up to 3.2 more on the total

Why it scored 84: Graded on the hosted REST API. Statuspage at status.belvo.com with components for the API, Widget, Webhooks, Dashboard, Recurrent Links, the sandbox and payments (20). The incident feed lists nothing between 10 July and 8 October 2026. Its newest entry is a minor one on 1 June, after a critical incident of 2.7 hours on 19 May and a major one of 10.9 hours on 15 May, both outside the 90 days (30). Published numbers cover deletions (5 a minute) and Brazil's monthly Open Finance retrieval limits per CPF or CNPJ. The general limits behind the 429 response have no numbers (8 of 15). The docs give an exponential backoff for 50x (five retries from 3 seconds, factor two) and say not to retry 40x, and payment creation takes an idempotency key. No Retry-After was found (12 of 15). Annex I of the terms is a service level agreement with support response times of 24 hours by priority, but it has no availability figure and the terms say uninterrupted availability isn't guaranteed (4 of 10). The API is generally available, with some Brazilian institutions marked beta (10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):

Hosted APIs, MCP servers, models and platforms.

- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).
- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.
- 15, rate limits documented with numbers.
- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.
- 10, an SLA published for any paid tier.
- 10, the surface agents use is generally available, not beta or preview.

Local packages, SDKs, frameworks and stdio MCP servers.

- 20, installs from an official package with supported runtimes stated.
- 25, a public CI and test suite, passing on the default branch.
- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).
- 15, semver discipline and breaking changes called out in a changelog.
- 15, version 1.0 or later, or declared stable.

Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.

## 7. Transparency & trust, 64 out of 100, up to 3.2 more on the total

Made of editorial 46, provenance 81.

Why it scored 64: Closed service with public terms, the General Conditions of Service Usage last updated 10 January 2025. The old npm package is MIT (15 of 30). The terms promise deletion within 30 days of termination or request and point to the retention controls in the docs, which set `stale_in`, `credentials_storage` and `save_data`. There are separate privacy policies for end users and for clients. The client policy gives retention of up to 10 years for contract records. No data processing agreement was found on the pages read (20 of 30). Fields are marked deprecated in the specification with no removal dates or notice period, and the terms give 30 days' notice for price changes only (5 of 20). The security page names AWS as host. No subprocessor list or data location was found, and the trust centre at trust.belvo.com is drawn by script and wasn't read (6 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 7 of the 7 things a reader expects, and has 1 clause that costs points (8 of 10)
- Privacy policy: read, states 5 of the 8 things a reader expects (7.8 of 10)
- Changelog: not found (0 of 10)
- security.txt: published but past its Expires date (5 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.

- The batch lead says balances and transactions in Brazil and Mexico. The docs and the MCP page read on 8 October 2026 show bank data for Brazil only, with employment, fiscal and direct debit products in Mexico
- The lead lists SDKs. The Python, Node and Ruby packages were last published in 2023 and github.com/belvo-finance/belvo-python, belvo-js and belvo-ruby return 404. Only the iOS and Android SDKs for biometric Pix are in the current docs
- The clean 90 days on the status page may mean no incidents or no posting. The feed can't tell the two apart
- Whether Belvo Instituição de Pagamento Ltda. is authorised by Banco Central do Brasil, and under what number. We found the name and CNPJ in the terms but didn't check the regulator's register
- unchecked: the trust centre at trust.belvo.com, which is drawn by script, so the subprocessor list and audit reports weren't read
- unchecked: the dashboard's activity logs and whether more than one key pair can exist per environment
- unchecked: the Data Access MCP's URL, tools and credential model, which have no public docs
- unchecked: the separate institutions status page that the docs mention
- unchecked: GitHub organisation listing and stars, because the GitHub API rate limit was reached

## Weaknesses

- Production keys follow a sales meeting, a certification call with a Belvo engineer and a paid plan from 1,000 USD a month
- One `secretId` and `secretPassword` pair per environment over HTTP Basic, with no scopes or read-only keys found
- No changelog was found, and the Python, Node and Ruby packages were last published in 2023 with their GitHub repositories returning 404
- Bank data covers Brazil only in the docs read on 8 October 2026, and Mexico has employment, fiscal and direct debit products
- security.txt expired on 22 March 2025
- General rate limits carry no published numbers, and the data webhooks have no signature, only an optional fixed authorisation header

## 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 HTTP Basic with `secretId` as username and `secretPassword` as password to https://sandbox.belvo.com, and switch to https://api.belvo.com only with production keys
- Create links through the Hosted Widget with a token from POST /api/token/, which expires after 10 minutes
- Read stored data with GET list calls. Each POST retrieval in Brazil counts against monthly Open Finance limits, such as 8 a month per CPF for owners and accounts
- Retry 50x responses up to five times from a 3 second base with a factor of two, and don't retry 40x apart from `too_many_sessions`
- Send `Belvo-Idempotency-Key` with a UUID when creating a payment intent, and `X-Belvo-Request-Mode: async` on link deletion to avoid the limit of 5 deletions a minute

## 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.