Browserless

# Fix list: Browserless

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/browserless, the October 2026 research run, assessed 8 October 2026. Grade BB, 70.4 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 Browserless: 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, 53 out of 100, up to 8.2 more on the total

Why it scored 53: An account holds up to 20 named API tokens, each shown once and revocable on its own, with a rotation guide. The hosted MCP server also accepts OAuth sign-in with PKCE. Cloud tokens have no scopes, and token roles apply only to self-hosted Enterprise containers (22). The docs give `?token=` in the URL as the standard way to authenticate REST and WebSocket calls and as an option for MCP, a secret in a query string (minus 10). The connector endpoint limits an assistant to five tools, account tools are read-only, the agent prompt tells the model to confirm controls marked destructive, and the draft Stripe Link checkout needs the buyer's approval for each purchase up to $50. No read-only token (10). Pages are untrusted content. Saved logins and card details are kept out of the model's context, but no prompt-injection guidance was found in the docs or the MCP agent prompt (3). Request logs record every request with endpoint, region, duration, error and cost, filterable by API key, and sessions can be replayed (13). SOC 2 Type II, third-party penetration tests and a disclosure policy with a 72-hour acknowledgement target on the trust centre. No security.txt and no bug bounty found (15).

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. Payments & pricing, 40 out of 100, up to 7.5 more on the total

Why it scored 40: No x402, MPP or L402 on any endpoint. The Stripe Link checkout tools let an agent pay a merchant and are not a way to pay Browserless (0). Plan prices, included units and per-unit overage rates of $0.0015 to $0.0020 are published without login, with the unit defined as up to 30 seconds of browser time (20). Free plan with 1,000 units a month, and the pricing page says no card is required (20). A person has to sign up in a browser to get a token or complete OAuth (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. Reliability, 82 out of 100, up to 3.6 more on the total

Why it scored 82: Graded as a hosted service, on the cloud API and its hosted MCP server. status.browserless.io runs on Better Stack with eight monitored components and 90 days of daily history (20). The page lists no incident reports for August to October 2026. Its monitors record 129 minutes of downtime on Unit and Usage Services on 11 September 2026, 45 minutes on the Account API on 5 September, and 3 minutes on US East, with US West, London and Amsterdam at 100 per cent, so minor only (20). Concurrency and session-length limits are published per plan, 2 to 100 browsers and 2 to 60 minutes, and the docs say there is no hourly or daily unit limit (15). 429 is documented as a full concurrency slot or queue, with a retry and exponential backoff guide and advice to retry only operations that are safe to run twice. No Retry-After header or idempotency key was found in the reviewed documentation (12). One docs page claims a 99.9% uptime SLA and the site lists SLAs under Enterprise, but no SLA terms are published (5). The REST, WebSocket and MCP surfaces are generally available, with the MCP server at 1.36.0 (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. Agent ergonomics, 78 out of 100, up to 3.6 more on the total

Why it scored 78: The full MCP server documents 14 tools (15), and a separate connector endpoint exposes five. The `browserless_agent` description is long, so we added back 5 of a possible 10 (20). `browserless_logs` pages by cursor, search, crawl and profiles take `limit`, crawl output is truncated per page, and Smart Scrape returns only the formats asked for (17). Errors come with a documented table of causes and fixes, a `requestId` that can be looked up in request logs, and `x-response-code` headers for the target site's status (16). Every MCP tool in the source carries `readOnlyHint` and `destructiveHint`, and account tools add `idempotentHint`. No idempotency keys, and the docs warn that a retried mutation can repeat a click or form submission (15). Few required parameters and existing Puppeteer or Playwright scripts connect by changing the endpoint, but the only published SDK is TypeScript. The docs say the Python SDK is not on PyPI yet (10).

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

- 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries).
- 20, pagination, filtering and output-size controls.
- 20, actionable, documented error responses, codes and messages an agent can recover from.
- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.
- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.

Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs.

## 5. Transparency & trust, 62 out of 100, up to 3.3 more on the total

Made of editorial 47, provenance 77.

Why it scored 62: The core server and MCP server are source-available under SSPL 1.0, which is not an OSI licence, or a commercial licence, and the cloud additions are closed under published terms (18). The privacy policy says Browserless does not store page content and keeps only an email address, and names Stripe, SendGrid, DigitalOcean and Fathom. The trust centre, updated 11 September 2026, says request logs, session replays, saved profiles and crawl results are stored in the United States on AWS, DigitalOcean and Supabase, and the pricing page gives replay storage of 1 to 90 days. The two do not agree, the policy is undated, and a DPA is on Enterprise plans only (12). No deprecation policy. The terms allow the service to be changed or discontinued without notice, though the docs flag the v1 API and the chrome.browserless.io endpoint as deprecated (5). Execution regions and storage locations are stated in detail. The subprocessor list is on request by email (12).

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):

- Domain age: browserless.io, registered 2017-09-06 (9 years) (11 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 3 of the 8 things a reader expects (6.3 of 10)
- security.txt: not found (0 of 10)

## 6. Schema & documentation, 89 out of 100, up to 1.8 more on the total

Why it scored 89: OpenAPI 3.0.0 at www.browserless.io/swagger.json with 87 paths and 100 operations, and Zod-typed inputs on every MCP tool in the source (25). llms.txt, llms-full.txt, a Markdown twin of every docs page, an API catalogue and an MCP server card under /.well-known (10). MCP descriptions state purpose and point to the right neighbour, for example `browserless_smartscraper` sends multi-page work to `browserless_crawl`. The agent tool's description is a long operating prompt (16). Inputs use enums and limits, but `browserless_function` takes free-form Puppeteer code and `browserless_agent` advertises a slimmed schema that the server re-validates at run time (12). Every OpenAPI operation lists 400, 401, 403, 404, 408, 429, 500 and 503, the docs carry an error table with fixes, and examples come in cURL, JavaScript and Python (13). A monthly product changelog and semver changelogs for the core (2.57.0) and MCP (1.36.0) repositories. REST paths carry no version (13).

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. Maintenance & community, 82 out of 100, up to 1.6 more on the total

Why it scored 82: @browserless.io/mcp 1.36.0 was published on 5 October 2026 and the core server 2.57.0 on 30 September (30). npm lists 32 stable MCP releases and 14 core releases since 10 July 2026, and the product changelog has an entry for each month from March to October 2026 (20). Both repositories show commits on 6 and 7 October 2026 and use release automation. GitHub's API refused our requests, so issue reply times went unread, and we scored the public changelogs and the support email only (12). The MCP server and the BAP TypeScript SDK (2.0.0) are current on npm, the Python SDK is unpublished, and the official MCP registry holds only a third-party entry for Browserless, so we scored the SDK line for an API (12). Test, lint, CodeQL and dependency-update workflows are in the repositories. We did not read their run results (8).

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

- 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older.
- 20, at least three releases or dated changelog entries in the last 90 days.
- 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15.
- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).
- 10, package health, current dependencies and CI.

Models are read for deprecation notice periods and model churn rather than release counts.

## What we couldn't check

What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it.

- unchecked: GitHub issue counts and reply times. GitHub's API refused our requests for the rate limit, and the issues page did not give a count to our reader
- unchecked: published security advisories for browserless/browserless. The advisories page did not render its list for our reader
- unchecked: whether CI passes on the default branches. We read the workflow files, not their results
- unchecked: the subprocessor list, which the trust centre says is available on request by email
- No SLA terms were found. One docs page claims a 99.9% uptime SLA and the site lists SLAs under Enterprise
- No Retry-After header, request-rate limit or idempotency key was found in the reviewed documentation
- The status page lists no incident reports, though its monitors recorded downtime on 5 and 11 September 2026. We did not establish what those outages affected for customers
- The source registers two Stripe Link tools beyond the 14 documented. The docs say they are a draft preview not yet enabled on the hosted fleet, so toolCount is 14

## Weaknesses

- REST and WebSocket routes take the API token as `?token=` in the URL, which the docs give as the standard method
- Cloud tokens have no scopes or roles. Token roles exist only on self-hosted Enterprise containers
- No x402, MPP or L402, and an account needs a person to sign up in a browser
- No prompt-injection guidance found in the docs or in the MCP agent prompt
- The terms allow the service to be changed or discontinued without notice, and the subprocessor list is on request only

## 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 the token as `Authorization: Bearer` to the MCP server. On REST and WebSocket routes it goes in `?token=`, so keep those URLs out of logs
- Close sessions as soon as the work is done. Browser time bills one unit per 30 seconds, rounded up, idle or not
- For multi-step work with `browserless_agent`, set `keepSessionAlive: true` on the first call and pass the returned `sessionId` after that. Sessions are one-shot by default since server 1.33.0
- On 429, back off and retry. It means the plan's concurrency limit or queue is full, usually from sessions left open
- Match the path to the protocol. Use `/chromium` for CDP clients and `/chromium/playwright` only for `playwright.connect()`

## 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.