Langfuse API + MCP by Langfuse (ClickHouse)

# Fix list: Langfuse API + MCP

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/langfuse, the October 2026 research run, assessed 1 October 2026. Grade BB, 72.8 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 Langfuse API + MCP: 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, 40 out of 100, up to 7.5 more on the total

Why it scored 40: No x402 or other machine payment (0). Per-unit pricing published without login, $8 per 100,000 units above the plan (20). Hobby is free with 50,000 units a month and no card (20). A person signs up in a browser to create keys. Free self-hosting isn't an agent route (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. Security & auth, 65 out of 100, up to 6.1 more on the total

Why it scored 65: Project-scoped key pairs over Basic auth, revocable, with organisation keys kept to admin APIs. No read-only key type (22). Annotations mark reads and destructive tools, but the server has no read-only mode and write tools are on by default (10). Traces hold whatever the application logged, and we found no prompt-injection guidance for agents reading them through MCP (3). Audit logs on Enterprise only (10). Bug bounty through ClickHouse's Bugcrowd programme, SOC 2 Type II, ISO 27001, an annual penetration test and GitHub advisories published in public. No security.txt (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.

## 3. Agent ergonomics, 74 out of 100, up to 4.2 more on the total

Why it scored 74: About 89 MCP tool definitions in the source on 1 October, all enabled by default, with no server-side toolsets. The docs say to restrict them with a client allowlist (5). Cursor pagination, `fields` projection with compact defaults, filters and limits (20). 429s carry `Retry-After`, and invalid MCP calls return named errors (16). 49 tools set `readOnlyHint: true` and 34 set `destructiveHint`. Dataset writes are upserts (18). Few required parameters. 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.

## 4. Reliability, 80 out of 100, up to 4 more on the total

Why it scored 80: incident.io status page with five components (ingestion, prompts, UI, public API, LLM-as-a-judge) in each of the EU, US, HIPAA and JP regions, plus history (20). Since 3 July the history lists 12 degraded-performance incidents, mostly ingestion and evaluation delays, among them 502 errors on the UI and APIs on 14 July and two on 25 August. None is marked as a full outage, but the count is high and durations weren't shown to us, so 15 rather than 20. Per-organisation rate limits published per bucket and plan (15). 429 with `Retry-After` documented as the authoritative wait (15). Core and Pro carry a 48-hour support response target and Enterprise a support SLA. We found no uptime SLA (5). API and MCP are GA (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. Schema & documentation, 93 out of 100, up to 1.1 more on the total

Why it scored 93: Public OpenAPI at cloud.langfuse.com, and every MCP tool takes a typed zod schema (25). llms.txt and Markdown docs (10). Tool descriptions say what to use each one for and how to size the result. `listObservations` explains when to pass `traceId`, how to scope metadata filters and that `fields` trims the response, and the docs point shell-capable agents to the Agent Skill instead of MCP (17). Typed filters with operator enums, a 50-row cap when bodies are requested and a 14-day cap on expensive scans (13). Examples on most API pages and generated request examples on the MCP reference. Error bodies are less fully documented (13). Dated changelog, a v2 API and a deprecation guide with a sunset date (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, 88 out of 100, up to 1.1 more on the total

Why it scored 88: Server v4.49.0 tagged on 2026-10-01 (30). Twelve server tags between 23 September and 1 October alone (20). New issues get triaged with labels within days. We couldn't see reply times, since GitHub's issue search is closed to our reader (15). Python SDK v4.16.0 on 2026-09-30 and JS SDK v5.11.1 on 2026-09-09 (15). CodeQL, Semgrep, Snyk and zizmor run in CI alongside the build pipeline. We couldn't confirm the pass state (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.

## 7. Transparency & trust, 87 out of 100, up to 1.1 more on the total

Made of editorial 91, provenance 82.

Why it scored 87: MIT core with the ee directories under a commercial licence (27). Data-retention docs, data regions, GDPR and HIPAA pages agree, though the privacy notice comes from ClickHouse, Inc., the imprint names Langfuse GmbH and the terms redirect to ClickHouse's general terms (24). Deprecated read APIs carry a published sunset date of 16 November 2026 with a migration guide (20). Self-hosted telemetry is documented field by field with `TELEMETRY_ENABLED=false` to turn it off, and the cloud has a live subprocessor list (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.

Provenance checks not met in full (half of this category, computed from checked facts):

- Domain age: langfuse.com, registered 2023-04-20 (3 years) (7 of 15)
- security.txt: not found (0 of 10)

## Deductions

Each comes off the total. A fixed and documented problem counts for less at the next check.

- 2025-11-09 to 2026-01-21. Three moderate advisories, cross-organisation enumeration of member and invitation lists (GHSA-94hf-6gqq-pj69), SSO account takeover through CSRF or phishing (GHSA-w9pw-c549-5m6w) and an unauthenticated Slack OAuth install that could link Slack to any project (GHSA-pvq7-vvfj-p98x). All fixed and published, eight to eleven months old, so -2 (https://github.com/langfuse/langfuse/security/advisories)

## 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.

- Incident durations on the status history weren't visible to our reader, so the severity of the 14 July and 25 August incidents is uncertain
- GitHub issue reply times and whether CI is passing on main, which GitHub's robots rules and our lack of API access kept us from checking

## Weaknesses

- About 89 MCP tools load by default, writes included, with no server-side toolsets or read-only mode
- 12 degraded incidents on the status page between 3 July and 30 September, mostly ingestion delays
- No uptime SLA found, only support response targets
- Hobby keeps 30 days of data and the General API bucket allows 30 requests a minute
- Self-hosting means running ClickHouse, Postgres, Redis and S3-compatible storage

## 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.

- Allowlist only the tools the agent needs. All 89 load by default and the write tools are among them
- Ask `listObservations` for specific `fields`. Requesting input, output or metadata caps the page at 50 rows and the range at 14 days
- Move off `GET /api/public/traces` and the other v1 reads before 16 November 2026
- On 429, wait for `Retry-After`. MCP calls share the organisation's General API bucket
- Pick the regional host (cloud, us.cloud, jp.cloud, hipaa.cloud) that matches the project's keys

## What the review panel asked for

- announcement dates on deprecations
- server-side toolsets
- fuller error bodies

## 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.