{
  "fixes": {
    "slug": "kiwi-mcp",
    "name": "Kiwi.com MCP Server",
    "listing": "https://www.anchorterminal.com/tools/kiwi-mcp",
    "markdown": "# Fix list: Kiwi.com MCP Server\n\nFrom Anchor Terminal's listing at https://www.anchorterminal.com/tools/kiwi-mcp, the October 2026 research run, assessed 8 October 2026. Grade D, 51.9 out of 100.\n\nThis 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.\n\nFor a coding agent working on Kiwi.com MCP Server: 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.\n\n## 1. Reliability, 45 out of 100, up to 11 more on the total\n\nWhy it scored 45: status.kiwi.com is a Statuspage site with components for Search, Search API, Booking and payments, and none for the MCP server (15 of 20). Its incident feed's newest entry is dated 13 October 2021, so the last 90 days are clean on paper, but five years without an entry on a large travel site is weak evidence of a kept record (20 of 30). No rate limits found in the README or the tool definitions (0). No 429, Retry-After or backoff guidance found (0). No SLA (0). Nothing calls the server beta, although `serverInfo.version` is empty and the README's VS Code snippet says 0.0.1 (10).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):\n\nHosted APIs, MCP servers, models and platforms.\n\n- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).\n- 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.\n- 15, rate limits documented with numbers.\n- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.\n- 10, an SLA published for any paid tier.\n- 10, the surface agents use is generally available, not beta or preview.\n\nLocal packages, SDKs, frameworks and stdio MCP servers.\n\n- 20, installs from an official package with supported runtimes stated.\n- 25, a public CI and test suite, passing on the default branch.\n- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).\n- 15, semver discipline and breaking changes called out in a changelog.\n- 15, version 1.0 or later, or declared stable.\n\nProtocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.\n\n## 2. Security \u0026 auth, 54 out of 100, up to 8.1 more on the total\n\nWhy it scored 54: No credentials at all. There is no secret to leak, and no way for an operator to scope, rotate or revoke access. `https://mcp.kiwi.com/.well-known/oauth-protected-resource` returns 404 (20 of 30). Search is read-only and nothing can spend money. The one write, `feedback-to-devs`, sends up to 3,000 characters of free text to Kiwi.com with no confirmation step (14 of 20). Results are structured fare data from Kiwi.com's own index with little free text. The tool description tells the model how to lay out results, to show the booking link and to close with a fun fact. No injection guidance found (8 of 15). No audit log or per-call view for an operator (0). security.txt with a contact, a disclosure policy that promises a reply in two working days, and a HackerOne bug bounty. No SOC 2 or ISO 27001 statement found on the pages read (12 of 20).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-security):\n\n- 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.\n- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.\n- 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.\n- 0 to 15, audit logs or per-call visibility for the operator.\n- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.\n\nModels 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.\n\n## 3. Maintenance \u0026 community, 13 out of 100, up to 7.6 more on the total\n\nWhy it scored 13: The only dated public change is the README repository's last commit on 7 August 2025, 14 months ago. The live server has changed since, because it now lists a second tool and filters the README doesn't describe, but nothing dates those changes (10 of 30). No releases or changelog entries in the last 90 days (0). Both issues on the public repository, opened 3 February and 9 September 2026, have no reply. `feedback-to-devs` is a support channel of a kind (3 of 15). Not in the official MCP registry. A search for kiwi returned five unrelated or third-party servers (0). No package, dependencies or CI to read (0).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):\n\n- 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.\n- 20, at least three releases or dated changelog entries in the last 90 days.\n- 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.\n- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).\n- 10, package health, current dependencies and CI.\n\nModels are read for deprecation notice periods and model churn rather than release counts.\n\n## 4. Schema \u0026 documentation, 55 out of 100, up to 7.3 more on the total\n\nWhy it scored 55: Both tools carry typed JSON Schema inputs and an output schema, read from `tools/list` on 8 October 2026 (25). No llms.txt on kiwi.com (404) and no Markdown docs beyond the README (0). The `search-flight` description states its purpose, when to call it and the result shape, and `feedback-to-devs` says when not to use it (15 of 20). 47 typed properties with ranges, two enums and three required fields, but dates are plain strings with no pattern, airline and airport lists are comma-separated strings, and the README says flexibility is up to 3 days where the schema allows 10 (11 of 15). Two example prompts in the README, no documented error responses (4 of 15). No changelog and no server version (0).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):\n\nAPIs and MCP servers.\n\n- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).\n- 10, llms.txt or Markdown docs served for agents.\n- 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.\n- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.\n- 0 to 15, examples and documented error responses.\n- 15, versioning and a public changelog.\n\nModels 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.\n\n## 5. Payments \u0026 pricing, 55 out of 100, up to 5.6 more on the total\n\nWhy it scored 55: No x402, MPP or L402 (0). The server cost nothing to call on 8 October 2026 and there is nothing to buy, but no page states that in writing, and fares include Kiwi.com's own charges without a published schedule (15 of 20). Free with no card and no account (20). Keyless, so an agent connects with no human signup (20).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):\n\nThe published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).\n\n- 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.\n- 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.\n- 20, a free tier or trial that doesn't need a card.\n- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).\n\nPayment 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.\n\nOpen-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.\n\n## 6. Agent ergonomics, 66 out of 100, up to 5.5 more on the total\n\nWhy it scored 66: Two tools, but their definitions total 19.3 KB of JSON because `search-flight` has 47 parameters and display instructions in its description (20 of 25). Filters cover price, stops, airlines, times, layovers, bags, weekdays and five sort orders, with no result limit or pagination. Two searches each returned 15 itineraries, 25 KB and 34 KB of text, with the same data again in `structuredContent` (10 of 20). A rejected input returned a raw Pydantic validation message in an `error` field with `isError` false, and no error list is published (6 of 20). `search-flight` is annotated read-only and idempotent, and `feedback-to-devs` not read-only and not destructive (20). Three required inputs, defaults for the rest and place names accepted. A date sent as 2026-11-12 was accepted although the schema asks for dd/mm/yyyy. No SDK, MCP only (10 of 15).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):\n\n- 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).\n- 20, pagination, filtering and output-size controls.\n- 20, actionable, documented error responses, codes and messages an agent can recover from.\n- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.\n- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.\n\nModels 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.\n\n## 7. Transparency \u0026 trust, 66 out of 100, up to 3 more on the total\n\nMade of editorial 44, provenance 87.\n\nWhy it scored 66: The server is closed source. The public repository holds one README and no licence file, and no terms name the MCP server, so the terms of use for the website and the booking terms and conditions are the nearest documents (10 of 30). The privacy policy, effective 7 September 2026, names Kiwi.com s.r.o. as controller and a data protection officer, lists data categories and purposes and gives a general retention of three years plus one. Its stated scope is the website and the mobile apps, and it doesn't mention MCP queries (20 of 30). No deprecation policy or dated notices (0). Infrastructure processors are named (Amazon Web Services, OVH, Hetzner, DigitalOcean, Akamai, Google Ireland) and transfers outside the EEA are described, without a full location list (14 of 20).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):\n\n- 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.\n- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).\n- 0 to 20, a deprecation policy or notices with dates.\n- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).\n\nThe other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.\n\nProvenance checks not met in full (half of this category, computed from checked facts):\n\n- Terms of service: read, states 3 of the 7 things a reader expects (6.6 of 10)\n- Changelog: not found (0 of 10)\n\n## What we couldn't check\n\nWhat 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.\n\n- When the live server last changed. It lists a tool and filters the README of 7 August 2025 doesn't describe, and nothing public dates them\n- Rate limits. None are published and we made six requests, too few to meet one\n- Whether Kiwi.com or Alpic operates the endpoint. The README repository belongs to alpic-ai and names Kiwi.com as author, and responses come from a uvicorn server behind Google and Fastly on mcp.kiwi.com\n- Which terms govern the MCP server. None name it, so `provenance.terms` points at the booking terms and conditions that apply once a user follows a booking link\n- Whether booking links carry a referral or pricing difference. `https://kiwi.com/u/\u003cid\u003e` redirects to www.kiwi.com and we didn't follow it further\n- unchecked: a launch announcement. A search of media.kiwi.com for MCP showed no result to our reader\n- unchecked: SOC 2, ISO 27001 or PCI statements. None on the security, privacy or terms pages we read, and we looked no further\n- The batch lead said one tool. The server lists two, `search-flight` and `feedback-to-devs`\n\n## Weaknesses\n\n- Search only. Booking, payment, changes and cancellations happen on kiwi.com through the `bookingUrl` in each result\n- No rate limits, 429 guidance, SLA, changelog or deprecation policy found, and `serverInfo.version` is empty\n- No result limit or pagination. Two searches returned 15 itineraries each, 25 KB and 34 KB of text, repeated in `structuredContent`\n- Validation failures come back as a raw Pydantic message in an `error` field with `isError` false\n- No terms name the MCP server, and the privacy policy's stated scope is the website and mobile apps\n- The public repository is a README last changed on 7 August 2025, with no licence, and it omits the second tool and most filters\n\n## What costs an agent a turn today\n\nThe notes we give agents before they call it. Each one is a workaround an agent shouldn't need.\n\n- POST to `https://mcp.kiwi.com` itself. `/mcp` returns 404, and responses arrive as `text/event-stream` even for a single reply\n- Send dates as dd/mm/yyyy and set `currency`, which defaults to EUR\n- Check the `error` field of every result. A rejected input returns `resultsCount` 0 with `isError` false\n- Narrow with `price_to`, `max_sector_stopovers` and `sort`. There is no limit parameter and each search returns 15 itineraries\n- Give the user `bookingUrl` to book. Don't call `feedback-to-devs` with user data, because it sends free text to Kiwi.com\n\n## When it's done\n\nSend 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.\n",
    "grade": "D",
    "score": 51.9,
    "assessed": "2026-10-08",
    "run": "October 2026 research run",
    "categories": [
      {
        "key": "reliability",
        "name": "Reliability",
        "score": 45,
        "maxGain": 11,
        "reason": "status.kiwi.com is a Statuspage site with components for Search, Search API, Booking and payments, and none for the MCP server (15 of 20). Its incident feed's newest entry is dated 13 October 2021, so the last 90 days are clean on paper, but five years without an entry on a large travel site is weak evidence of a kept record (20 of 30). No rate limits found in the README or the tool definitions (0). No 429, Retry-After or backoff guidance found (0). No SLA (0). Nothing calls the server beta, although `serverInfo.version` is empty and the README's VS Code snippet says 0.0.1 (10).",
        "checklist": [
          "Hosted APIs, MCP servers, models and platforms.",
          "- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).\n- 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.\n- 15, rate limits documented with numbers.\n- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.\n- 10, an SLA published for any paid tier.\n- 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.\n- 25, a public CI and test suite, passing on the default branch.\n- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).\n- 15, semver discipline and breaking changes called out in a changelog.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-reliability"
      },
      {
        "key": "security",
        "name": "Security \u0026 auth",
        "score": 54,
        "maxGain": 8.1,
        "reason": "No credentials at all. There is no secret to leak, and no way for an operator to scope, rotate or revoke access. `https://mcp.kiwi.com/.well-known/oauth-protected-resource` returns 404 (20 of 30). Search is read-only and nothing can spend money. The one write, `feedback-to-devs`, sends up to 3,000 characters of free text to Kiwi.com with no confirmation step (14 of 20). Results are structured fare data from Kiwi.com's own index with little free text. The tool description tells the model how to lay out results, to show the booking link and to close with a fun fact. No injection guidance found (8 of 15). No audit log or per-call view for an operator (0). security.txt with a contact, a disclosure policy that promises a reply in two working days, and a HackerOne bug bounty. No SOC 2 or ISO 27001 statement found on the pages read (12 of 20).",
        "checklist": [
          "- 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.\n- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.\n- 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.\n- 0 to 15, audit logs or per-call visibility for the operator.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-security"
      },
      {
        "key": "maintenance",
        "name": "Maintenance \u0026 community",
        "score": 13,
        "maxGain": 7.6,
        "reason": "The only dated public change is the README repository's last commit on 7 August 2025, 14 months ago. The live server has changed since, because it now lists a second tool and filters the README doesn't describe, but nothing dates those changes (10 of 30). No releases or changelog entries in the last 90 days (0). Both issues on the public repository, opened 3 February and 9 September 2026, have no reply. `feedback-to-devs` is a support channel of a kind (3 of 15). Not in the official MCP registry. A search for kiwi returned five unrelated or third-party servers (0). No package, dependencies or CI to read (0).",
        "checklist": [
          "- 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.\n- 20, at least three releases or dated changelog entries in the last 90 days.\n- 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.\n- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).\n- 10, package health, current dependencies and CI.",
          "Models are read for deprecation notice periods and model churn rather than release counts."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-maintenance"
      },
      {
        "key": "schema",
        "name": "Schema \u0026 documentation",
        "score": 55,
        "maxGain": 7.3,
        "reason": "Both tools carry typed JSON Schema inputs and an output schema, read from `tools/list` on 8 October 2026 (25). No llms.txt on kiwi.com (404) and no Markdown docs beyond the README (0). The `search-flight` description states its purpose, when to call it and the result shape, and `feedback-to-devs` says when not to use it (15 of 20). 47 typed properties with ranges, two enums and three required fields, but dates are plain strings with no pattern, airline and airport lists are comma-separated strings, and the README says flexibility is up to 3 days where the schema allows 10 (11 of 15). Two example prompts in the README, no documented error responses (4 of 15). No changelog and no server version (0).",
        "checklist": [
          "APIs and MCP servers.",
          "- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).\n- 10, llms.txt or Markdown docs served for agents.\n- 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.\n- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.\n- 0 to 15, examples and documented error responses.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-schema"
      },
      {
        "key": "payments",
        "name": "Payments \u0026 pricing",
        "score": 55,
        "maxGain": 5.6,
        "reason": "No x402, MPP or L402 (0). The server cost nothing to call on 8 October 2026 and there is nothing to buy, but no page states that in writing, and fares include Kiwi.com's own charges without a published schedule (15 of 20). Free with no card and no account (20). Keyless, so an agent connects with no human signup (20).",
        "checklist": [
          "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.\n- 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.\n- 20, a free tier or trial that doesn't need a card.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-payments"
      },
      {
        "key": "ergonomics",
        "name": "Agent ergonomics",
        "score": 66,
        "maxGain": 5.5,
        "reason": "Two tools, but their definitions total 19.3 KB of JSON because `search-flight` has 47 parameters and display instructions in its description (20 of 25). Filters cover price, stops, airlines, times, layovers, bags, weekdays and five sort orders, with no result limit or pagination. Two searches each returned 15 itineraries, 25 KB and 34 KB of text, with the same data again in `structuredContent` (10 of 20). A rejected input returned a raw Pydantic validation message in an `error` field with `isError` false, and no error list is published (6 of 20). `search-flight` is annotated read-only and idempotent, and `feedback-to-devs` not read-only and not destructive (20). Three required inputs, defaults for the rest and place names accepted. A date sent as 2026-11-12 was accepted although the schema asks for dd/mm/yyyy. No SDK, MCP only (10 of 15).",
        "checklist": [
          "- 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).\n- 20, pagination, filtering and output-size controls.\n- 20, actionable, documented error responses, codes and messages an agent can recover from.\n- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-ergonomics"
      },
      {
        "key": "transparency",
        "name": "Transparency \u0026 trust",
        "score": 66,
        "maxGain": 3,
        "reason": "The server is closed source. The public repository holds one README and no licence file, and no terms name the MCP server, so the terms of use for the website and the booking terms and conditions are the nearest documents (10 of 30). The privacy policy, effective 7 September 2026, names Kiwi.com s.r.o. as controller and a data protection officer, lists data categories and purposes and gives a general retention of three years plus one. Its stated scope is the website and the mobile apps, and it doesn't mention MCP queries (20 of 30). No deprecation policy or dated notices (0). Infrastructure processors are named (Amazon Web Services, OVH, Hetzner, DigitalOcean, Akamai, Google Ireland) and transfers outside the EEA are described, without a full location list (14 of 20).",
        "blend": "editorial 44, provenance 87",
        "checklist": [
          "- 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.\n- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).\n- 0 to 20, a deprecation policy or notices with dates.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-transparency"
      }
    ],
    "provenance": [
      {
        "label": "Terms of service",
        "value": "read, states 3 of the 7 things a reader expects",
        "points": 6.6,
        "max": 10
      },
      {
        "label": "Changelog",
        "value": "not found",
        "points": 0,
        "max": 10
      }
    ],
    "unchecked": [
      "When the live server last changed. It lists a tool and filters the README of 7 August 2025 doesn't describe, and nothing public dates them",
      "Rate limits. None are published and we made six requests, too few to meet one",
      "Whether Kiwi.com or Alpic operates the endpoint. The README repository belongs to alpic-ai and names Kiwi.com as author, and responses come from a uvicorn server behind Google and Fastly on mcp.kiwi.com",
      "Which terms govern the MCP server. None name it, so `provenance.terms` points at the booking terms and conditions that apply once a user follows a booking link",
      "Whether booking links carry a referral or pricing difference. `https://kiwi.com/u/\u003cid\u003e` redirects to www.kiwi.com and we didn't follow it further",
      "unchecked: a launch announcement. A search of media.kiwi.com for MCP showed no result to our reader",
      "unchecked: SOC 2, ISO 27001 or PCI statements. None on the security, privacy or terms pages we read, and we looked no further",
      "The batch lead said one tool. The server lists two, `search-flight` and `feedback-to-devs`"
    ],
    "weaknesses": [
      "Search only. Booking, payment, changes and cancellations happen on kiwi.com through the `bookingUrl` in each result",
      "No rate limits, 429 guidance, SLA, changelog or deprecation policy found, and `serverInfo.version` is empty",
      "No result limit or pagination. Two searches returned 15 itineraries each, 25 KB and 34 KB of text, repeated in `structuredContent`",
      "Validation failures come back as a raw Pydantic message in an `error` field with `isError` false",
      "No terms name the MCP server, and the privacy policy's stated scope is the website and mobile apps",
      "The public repository is a README last changed on 7 August 2025, with no licence, and it omits the second tool and most filters"
    ],
    "agentNotes": [
      "POST to `https://mcp.kiwi.com` itself. `/mcp` returns 404, and responses arrive as `text/event-stream` even for a single reply",
      "Send dates as dd/mm/yyyy and set `currency`, which defaults to EUR",
      "Check the `error` field of every result. A rejected input returns `resultsCount` 0 with `isError` false",
      "Narrow with `price_to`, `max_sector_stopovers` and `sort`. There is no limit parameter and each search returns 15 itineraries",
      "Give the user `bookingUrl` to book. Don't call `feedback-to-devs` with user data, because it sends free text to Kiwi.com"
    ],
    "recheck": "https://www.anchorterminal.com/builders/#disputes"
  },
  "meta": {
    "attribution": "Anchor Terminal (https://www.anchorterminal.com)",
    "docs": "https://www.anchorterminal.com/docs/",
    "generatedAt": "2026-10-08",
    "license": "CC-BY-4.0",
    "method": "https://www.anchorterminal.com/benchmark/",
    "methodology": "0.4",
    "openapi": "https://www.anchorterminal.com/openapi.json",
    "preview": false,
    "run": "2026-10-01",
    "runLabel": "October 2026 research run"
  }
}