Airtop

# Fix list: Airtop

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/airtop, the October 2026 research run, assessed 8 October 2026. Grade C, 55.3 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 Airtop: 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, 46 out of 100, up to 9.5 more on the total

Why it scored 46: API keys from the portal, sent as a Bearer header on REST, CDP and Selenium, so no secret travels in a URL. Scopes, expiry and rotation were not found in the reviewed documentation, and the portal's key page sits behind login (18). No read-only keys or approval step. A person can log in through Live View so the agent never holds the password, profiles are encrypted at rest with AES-256, each browser runs in its own virtual machine per the docs, and credit thresholds cap a call (8). The API returns untrusted page content and runs models over it, and no prompt-injection guidance was found (0). Optional session recordings, a sessions page with logs in the portal, and credit usage and a request ID on every AI response (10). Airtop states SOC 2 Type II and HIPAA compliance with the report on request, and links a Vanta trust centre that did not render for us. No security.txt, disclosure policy or bug bounty was found (10).

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, 25 out of 100, up to 9.4 more on the total

Why it scored 25: No x402, MPP or L402, and an unauthenticated request returned 401 (0). Plan prices are public, $0, $29, $189 and $558 a month with stated credit allowances, but the credits charged for browser time, proxy traffic and AI calls are not published, so plan-only (10). The Free plan renews 1,000 credits a month. Whether signup asks for a card is not stated on the pricing page or in the docs, so we award 15 of 20. A person has to sign up in a browser to get a key (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. Maintenance & community, 24 out of 100, up to 6.7 more on the total

Why it scored 24: Graded on the browser API. The newest published artefact is the Node SDK's 1.0.0-alpha2.47 on 1 July 2026, 99 days before the check. The stable 0.1.47 dates from 6 November 2025 and the docs changelog ends on 3 September 2025 (10). No three releases or dated entries in the last 90 days (0). Support is by Discord and email. On GitHub the Node SDK has an issue from February 2025 still open and six pull requests open since early 2025 (5). Official SDKs exist but the Python one was last released on 23 April 2025, and no Airtop server is in the MCP registry (6). Both SDK repositories carry CI workflows. Their results were not read (3).

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. Reliability, 69 out of 100, up to 6.2 more on the total

Why it scored 69: Graded as a hosted service, on the cloud browser API. status.airtop.ai runs on Instatus with 90 days of history, though it lists a single component, App (20). Between 10 July and 8 October 2026 it records no incident, only 30 minutes of scheduled maintenance on 18 August. A proxies incident of 5 hours 54 minutes on 1 July falls just outside the window (30). No request rate limit was found in the reviewed documentation. The pricing page gives simultaneous sessions per plan, 3, 30 and 100 (5). No 429 or backoff guidance and no idempotency keys were found. The SDK helpers retry twice by default and the CDP guides table a 503 for sessions that cannot be reached (4). No SLA found for any plan (0). The API is at /v1 and sessions and windows are not marked beta, though form filling is (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.

## 5. Agent ergonomics, 65 out of 100, up to 5.7 more on the total

Why it scored 65: AI calls return an answer shaped by the prompt and an optional `outputSchema`, with `costThresholdCredits` and `timeThresholdSeconds` to bound a call. Scrape returns the page content without field selection (18). Sessions list by `offset` and `limit` with a default of 10 and filters for status and IDs, files list the same way, and `paginated-extraction` follows a site's own pages (15). Responses share an envelope with `errors` and `warnings` carrying `message`, `code` and `reason`, and AI calls report success, partial or failure. The codes are not listed, and the 401 bodies we received had a message only (12). No idempotency keys. `clientRequestId` is echoed for correlation, and async variants return a request ID to poll (7). A session needs no parameters, and there are official Node and Python SDKs, the Python one last released in April 2025 (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.

## 6. Schema & documentation, 76 out of 100, up to 3.9 more on the total

Why it scored 76: OpenAPI 3.1 at docs.airtop.ai/openapi.json with 31 paths and 39 operations and a Bearer scheme (25). llms.txt and a Markdown twin of every docs page (10). 32 of 39 operations carry a description, parameter descriptions are long and specific, and the two deprecated endpoints name their replacement. Few say when not to use a call (13). Inputs carry 18 enums, required fields and bounds such as `timeoutMinutes` from 1 to 10,080, but `outputSchema` is a JSON Schema passed as a string and the proxy setting is loosely typed (11). The Markdown reference shows request and response examples. The spec declares error responses on 5 of 39 operations and no 429 anywhere, and no list of issue codes was found (8). Paths are versioned under /v1 and a dated changelog is public, but its last entry is 3 September 2025 (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, 61 out of 100, up to 3.4 more on the total

Made of editorial 42, provenance 80.

Why it scored 61: A closed service with published terms of use (effective 28 July 2025). The SDKs are MIT per npm and PyPI (17). The privacy policy (updated 2 September 2025) keeps personal data as long as necessary, with no periods, and the site and FAQ say customer data is not used for AI training, by Airtop or its model suppliers. No public DPA or retention period for session data, recordings or profiles was found (12). No deprecation policy. Deprecated fields and two deprecated endpoints are marked in the reference with replacements and no dates (5). The privacy policy says hosting is in the United States, and the FAQ names OpenAI, Anthropic and Google as model suppliers. No subprocessor list was read, since the trust centre did not render (8).

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: airtop.ai, registered 2024-07-08 (2 years) (7 of 15)
- Terms of service: read, states 6 of the 7 things a reader expects (9.1 of 10)
- Privacy policy: read, states 7 of the 8 things a reader expects (9.3 of 10)
- security.txt: not found (0 of 10)

## What we couldn't check

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

- unchecked: the contents of the Vanta trust centre at trust.airtop.ai (certifications, subprocessors, policies), which needs JavaScript and did not render for our reader
- unchecked: the signup flow at portal.airtop.ai, which redirects to a login host, so whether the Free plan asks for a card and how keys are named, revoked or rotated was not seen
- unchecked: CI results for the SDK repositories
- unchecked: the Google Drive copies of the terms and privacy policy linked from each page were not opened
- The credits charged for a browser minute, a GB of integrated proxy traffic and each AI call are not on the pricing page or in the docs
- The pricing page states a 10,000 one-time credit bonus on the Free plan while the billing guide states a 30,000 signup grant
- The lead's description is accurate, but Airtop's site now leads with Agent Builder and Mark, and the developer changelog has had no entry since 3 September 2025. Whether the browser API is still actively developed beyond the alpha Node SDK is not stated
- The 1.0.0-alpha2 Node SDK names a repository, github.com/airtop-ai/airtop-sdk, that is not among the organisation's public repositories
- The API's response headers name an mcp-connect.airtop.ai host, which answered 502. No product MCP server is documented
- The terms of use are the only agreement found. No separate API terms, DPA or SLA is published

## Weaknesses

- No published rate for what a browser minute, a GB of proxy traffic or an AI call costs in credits
- The developer changelog's last entry is 3 September 2025, and the stable Python SDK dates from 23 April 2025
- No request rate limits, 429 handling or idempotency keys found in the reviewed documentation
- No scoped or read-only API keys documented, and no guidance on prompt injection from page content
- No security.txt, bug bounty, public DPA or subprocessor list found. The SOC 2 report is on request

## 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.

- Terminate every session with `DELETE /v1/sessions/{id}`. Sessions bill in 30-second increments and idle ones last 10 minutes by default
- Set `costThresholdCredits` on AI calls and read `meta.usage.credits` afterwards. Credit rates are not published, so measure them
- Use `page-query` with `configuration.outputSchema` for structured output. `prompt-content` and `summarize-content` are deprecated
- When calling REST without the SDK, poll session status until `running` before creating a window. Only the SDK waits for you
- Airtop's integrated proxy cannot reach LinkedIn, ticketing sites and some banks and streaming services. Bring a custom proxy for those

## 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.