Cloudflare Web Search API

# Fix list: Cloudflare Web Search API

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/cloudflare-web-search, the October 2026 research run, assessed 5 October 2026. Grade B, 62.1 out of 100.

This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public.

For a coding agent working on Cloudflare Web Search API: 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 on /ai/websearch. AI Gateway Machine Payments covers only /ai/run for four open models (0). Per-search prices published without a login, $0.25, $5.00 and $7.00 per 1,000 by provider. The 5 per cent fee on credit purchases sits on the Unified Billing page, not the providers page (20). No free allowance found. Credits need a payment method, and the BYOK route needs an account with the provider (0). A person signs up for Cloudflare in a browser and buys credits or stores a provider 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.

## 2. Reliability, 55 out of 100, up to 9 more on the total

Why it scored 55: Scored as a hosted API. www.cloudflarestatus.com is a Statuspage with AI Gateway, Workers AI and API components (20). The incidents feed covered 17 September to 5 October, 50 incidents, none naming AI Gateway or Workers AI. The API component had a minor delay on membership permission changes on 30 September. The history page is paginated by script, so the rest of the 90 days went unread, and the product is three days old (8 of 30). 200 requests per 60 seconds per gateway on Unified Billing credentials, and 1,200 requests per five minutes per Cloudflare API token (15). The Cloudflare API returns 429 with `retry-after` and `Ratelimit` headers, and the error body carries a `retryable` flag. Searches have no side effects, but 429 isn't among the responses the OpenAPI operation lists (12 of 15). No SLA for AI Gateway found (0). The docs label it open beta, although the OpenAPI operation is tagged generally-available. We scored from the docs (0).

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.

## 3. Agent ergonomics, 63 out of 100, up to 6 more on the total

Why it scored 63: Results are small by default (URL, title, description), and `limit` caps them at 1 to 10. The default provider, Ceramic.ai, returns descriptions of up to 8,000 characters a result with no parameter to shorten them (15 of 25). `limit` is the only control. No pagination, offset, domain, date, country or freshness filters (6 of 20). Errors come back with `category` (gateway, credential, provider, internal), `code`, `status`, `retryable` and `gatewayRequestId`. The codes themselves aren't listed (16 of 20). Searches only read, the error body says whether a retry helps, and gateways can retry upstream failures up to five times. No idempotency keys, which a read doesn't need (16 of 20). Only `query` and the gateway ID are required. The Workers binding is TypeScript, and the Cloudflare TypeScript SDK 7.3.0 of 3 October has no web search method. We didn't check the Python or Go SDKs (10 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.

## 4. Security & auth, 71 out of 100, up to 5.1 more on the total

Why it scored 71: Cloudflare API tokens are scoped by permission and revocable, and web search needs only Workers AI Read and AI Gateway Read. Permission is per account, not per gateway or provider. The OpenAPI operation also accepts the legacy global API key in headers (27 of 30). Every call reads. Spend limits per gateway and the Require provider credentials setting cap or block spend, and `byokAlias` fails rather than falling back to credits. Read permission is enough to spend credits (14 of 20). Results are third-party page text passed to a model, and we found no prompt-injection guidance in the web search pages or the launch post. AI Gateway Guardrails aren't documented for web search (0 of 15). Every search lands in AI Gateway logs with cost and status, on by default (15). security.txt names HackerOne and a disclosure policy, read on 5 October. Certifications weren't checked this run (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.

## 5. Schema & documentation, 81 out of 100, up to 3.1 more on the total

Why it scored 81: Cloudflare's public OpenAPI has `/accounts/{account_id}/ai/websearch` with a typed body, `additionalProperties: false` and typed 200, 400, 403 and 502 responses. `provider` is a free string there where the docs give an enum (22 of 25). llms.txt for the product and Markdown for every page (10). The providers page says which provider suits which job, such as Ceramic.ai for agents that run many searches and Exa for short query-relevant excerpts. Nothing says when not to use the endpoint, and the OpenAPI description is one line (12 of 20). `query` 1 to 1,024 characters, `limit` 1 to 10, `provider` enum in the docs, `byokAlias` with a pattern (14 of 15). curl and Worker examples, a response example, and a typed error body with category, code and `retryable`. No catalogue of error codes or messages found (11 of 15). Dated launch entry in the AI Gateway changelog. The API path carries v4 for the whole Cloudflare API, nothing specific to web search (12 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.

## 6. Maintenance & community, 71 out of 100, up to 2.5 more on the total

Why it scored 71: Read for a closed service. Launched on 2 October 2026 (30). Ten dated AI Gateway changelog entries since 5 August, including Machine Payments on 30 September and auto-retry (20). A public changelog with RSS and a developer Discord. We didn't check support response times (8 of 15). The Workers binding has the method. The Cloudflare TypeScript SDK 7.3.0, published on 3 October, doesn't, though the OpenAPI already names an SDK method `websearch` (8 of 15). The OpenAPI spec carries the path a day after launch. Nothing else to measure for a hosted endpoint (5 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.

## 7. Transparency & trust, 75 out of 100, up to 2.2 more on the total

Made of editorial 50, provenance 100.

Why it scored 75: Closed service under Cloudflare's terms, with each provider's terms linked from the providers page (15 of 30). AI Gateway logs are on by default with headers to skip payloads, and Cloudflare's privacy policy and DPA exist per the Cloudflare MCP check. The providers page gives Zero Data Retention for Ceramic.ai and Linkup but not Exa, while the 2 October changelog says all three support it. No retention period for web search logs on the pages we read (15 of 30). No deprecation policy found. The changelog posts dated changes, such as the 24 September log cut-over, but no deprecation notices (6 of 20). The three providers are named, and Cloudflare's subprocessor page names 12 third parties with locations per the Cloudflare MCP check. Where searches are processed isn't stated (14 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.

## 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: incident history on cloudflarestatus.com before 17 September 2026 (the history page is paginated by script)
- Whether Exa searches through Cloudflare have Zero Data Retention. The providers page says no and the 2 October changelog says all three providers support it
- Whether failed or empty searches are charged, and whether a 502 provider failure draws credits
- When the endpoint leaves open beta. The docs say beta and the OpenAPI tags it generally-available
- unchecked: the Cloudflare Python and Go SDKs for a web search method
- unchecked: Cloudflare's certifications and how long AI Gateway keeps web search logs

## Weaknesses

- Open beta, with no SLA found for AI Gateway
- Credits need a payment method and carry a 5 per cent purchase fee. No free allowance
- At most 10 results a call, with no pagination, domain or date filters
- No x402. Machine Payments covers only /ai/run
- The changelog and the providers page disagree on whether Exa searches have Zero Data Retention

## 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.

- Set `options.gateway.id`. Every account has a gateway named `default`
- Pass `provider` explicitly. It defaults to Ceramic.ai, whose descriptions can run to 8,000 characters a result
- Set `limit` to the results you'll read, between 1 and 10
- Set `byokAlias` when you mean to use a stored provider key, so a missing key fails with 400 instead of spending credits
- Send `cf-aig-collect-log-payload: false` if queries shouldn't be stored in gateway logs

## What the review panel asked for

- State billing for failed searches
- Show fee on providers page
- date and domain filters
- a description length control

## 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.