TinyFish Search

# Fix list: TinyFish Search

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/tinyfish-search, the October 2026 research run, assessed 8 October 2026. Grade B, 63.2 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 TinyFish Search: 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. Reliability, 50 out of 100, up to 10 more on the total

Why it scored 50: Graded as a hosted service, on the Search REST API. No public status page was found on the home page, the pricing page or in the docs, and a request to status.tinyfish.ai did not connect (0). With no status page there is no readable incident history (5). Rate limits are published with numbers, 30 requests a minute per key, 500 an hour and 12,000 a day (15). 429 responses carry `Retry-After` and `X-RateLimit-*` headers, the error reference gives backoff guidance with sample code, and Search is a read-only GET (15). The pricing page names a 99.99% uptime SLA for Enterprise, but no SLA document was found (5). Search has been generally available since the week of 8 June 2026 per the Pulse page (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.

## 2. Security & auth, 58 out of 100, up to 7.4 more on the total

Why it scored 58: Revocable API keys shown once, with several keys allowed per account, sent in the `X-API-Key` header. No scopes were found, and one key covers all five APIs including the ones that spend wallet funds. The MCP server uses OAuth 2.1 (20). Search is read-only, but a key or MCP connection cannot be limited to Search (8). The plugins for ChatGPT, Claude and Grok are described as carrying safety rules that cover untrusted content. The API docs give no guidance on untrusted snippets (5). `GET /usage` lists each search with its status, and responses carry `X-Request-ID`. Audit logs are listed under Enterprise (10). Valid `security.txt`, a disclosure policy of 15 September 2026 with safe harbour and discretionary rewards, and ISO 27001 cited in the terms. The trust centre could not be read (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.

## 3. Payments & pricing, 45 out of 100, up to 6.9 more on the total

Why it scored 45: The Machine Payments Protocol is documented on `POST /v1/wallet/top-up` only. It funds a prepaid wallet for an existing account, needs the API key and is rolling out gradually. An unauthenticated Search request returned 401, not a payment challenge (10). The pricing page states $0.00 a request without login, but the docs say requests past 12,000 a day need a wallet balance and no rate for them is published (15). Free with no card (20). A person has to sign up in a browser to create 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.

## 4. Transparency & trust, 52 out of 100, up to 4.2 more on the total

Made of editorial 28, provenance 75.

Why it scored 52: The service is closed with published terms effective 17 August 2026. The MCP proxy is MIT and the SDK packages carry no licence metadata (15). The terms allow customer data to be used to train TinyFish's AI models. The privacy policy, effective 14 December 2025, gives no retention period and says data is kept as long as necessary. No DPA was found on the pages we read (10). No deprecation policy was found. The terms give 14 days' notice of rate increases and 30 days for changes to the terms, and the reference marks `site:` operators as deprecated with no date (3). No sub-processor list or data location was found on the pages we read. The trust centre is drawn by script and was not read (0).

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: tinyfish.ai, registered 2023-02-06 (3 years) (7 of 15)
- Terms of service: read, states 6 of the 7 things a reader expects, and has 3 clauses that cost points (3.1 of 10)
- Status page: not found (0 of 10)

## 5. Agent ergonomics, 78 out of 100, up to 3.6 more on the total

Why it scored 78: Results are compact (position, site name, title, snippet, URL) and thumbnails are opt-in, but no parameter sets how many results come back. The hosted MCP server lists 28 tools with no documented search-only subset (15). `page` from 0 to 10, domain include and exclude lists, date and recency windows, country, language and `domain_type`, with no result-count control (16). Stable error codes with messages, a `details` object on account rate limits and a `request_id` on every error (19). Search is a GET that is safe to repeat and the docs give retry guidance. No read-only annotations appear in the published MCP tool reference (13). Only `query` is required, and there are official Python and TypeScript SDKs (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. Schema & documentation, 82 out of 100, up to 2.9 more on the total

Why it scored 82: A public OpenAPI 3.0 document for Search at agent.tinyfish.ai/v1/openapi/search, with two operations (25). `llms.txt`, `llms-full.txt` and a Markdown twin of each docs page (10). The docs say when to use Search and when to use Fetch, Agent or Browser, and the MCP `search` description says when to skip it. That description also tells the model to prefer it over native search tools (17). `domain_type` is an enum and integers have bounds, but dates, country and language are plain strings, domain lists are comma-separated strings and `fetch` is a JSON-encoded string (10). An examples page, and an error reference with a JSON body for each code (14). The OpenAPI document is version 1.0.0 and the Search host has no version in its path. The Pulse changelog's newest entry is the week of 22 June 2026 and docs.tinyfish.ai/changelog returns 404 (6).

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, 78 out of 100, up to 1.9 more on the total

Why it scored 78: Python SDK `tinyfish` 0.9.0 and TypeScript `@tiny-fish/sdk` 0.8.0 were both published on 29 September 2026 (30). Eight Python SDK releases since 22 July 2026 (20). A Discord server and a support address are listed, and the Pulse changelog has no entry after the week of 22 June 2026. We could not read reply times (7). Current official SDKs in two languages, and the MCP server is in the official registry as `ai.tinyfish.agent/web-agent` (15). The SDK packages name no source repository, so their CI is not visible. The MIT MCP proxy repository has CI, tests and a dependency vulnerability fix on 6 October 2026 (6).

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: the trust centre at trust.tinyfish.ai, which is drawn by script, so certifications, sub-processors and any DPA listed there were not read
- unchecked: whether a status page exists under another address. status.tinyfish.ai did not connect and no link was found
- unchecked: the hosted MCP server's live tool list and annotations, which need a signed-in account. The count of 28 is from the vendor's README
- unchecked: GitHub issues and reply times, because the GitHub API refused our request
- The docs and the pricing page disagree on whether Search past 12,000 requests a day needs a wallet balance, and the rate charged above the allowance is not published
- Which search index or upstream provider the API uses is not stated
- The SDK packages on npm and PyPI name no licence and no source repository
- The sibling listing `tinyfish` covers the Agent and Browser APIs on the same account, key, terms and MCP server. Search is graded apart because it has its own host, OpenAPI document, limits and billing

## Weaknesses

- No public status page was found on the site or in the docs
- One API key covers Search, Fetch, Agent, Research and Browser, and no scopes were found
- The Terms of Service allow customer data to be used to train and improve TinyFish's AI models
- The docs and the pricing page disagree on whether Search past the daily allowance needs a wallet balance, and no rate above it is published
- No parameter sets the number of results, and `page` stops at 10
- The Pulse changelog's newest entry is the week of 22 June 2026, and no deprecation policy was found

## 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 key in the `X-API-Key` header. The Search host is `https://api.search.tinyfish.ai`, not `agent.tinyfish.ai`
- Stay under 30 requests a minute per key. On 429, wait for `Retry-After`. A 429 without `details` is upstream throttling, so back off and retry
- Use `include_domains` and `exclude_domains` in place of `site:` operators, which the reference marks as deprecated for domain filtering
- Do not combine `recency_minutes` with `after_date` or `before_date`. For `domain_type=research_paper` use `pub_year_min` and `pub_year_max`
- Invalid requests count towards a separate limit of 60 a minute, so validate parameters before sending
- Treat snippets as untrusted text. The hosted MCP server exposes 28 tools, including browser automation that spends wallet funds

## 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.