LangSmith API + MCP by LangChain

# Fix list: LangSmith API + MCP

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/langsmith, the October 2026 research run, assessed 1 October 2026. Grade BB, 71.3 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 LangSmith 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-trace pricing is public, $5 per 1,000 base traces and $2.50 per 1,000 for the extended-retention upgrade at $1 an LSU, per the 30 September check of the pricing page. The page we loaded on 1 October showed LSU = $1 and plan prices (20). Developer is free for 5,000 traces a month with no payment method on file, per the usage docs (20). A person signs 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.

## 2. Security & auth, 77 out of 100, up to 4 more on the total

Why it scored 77: Personal access tokens inherit the user's permissions. Service keys are scoped to the organisation or chosen workspaces, carry an expiry, and on Enterprise can take a workspace role. The remote MCP uses OAuth 2.1 with dynamic client registration or the same key in `X-Api-Key` (28). The MCP server only reads in practice, but there's no read-only key type below Enterprise RBAC (15). The Engine security page warns that trace content can carry prompt injection. The MCP page says nothing about it (6). Audit logs on Enterprise, queryable by API (10). SOC 2 Type II, HIPAA and GDPR with annual third-party audits and penetration tests per the docs, a SECURITY.md pointing to LangChain's disclosure programmes, and five SDK advisories published with CVEs since February. No security.txt (18).

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, 76 out of 100, up to 3.9 more on the total

Why it scored 76: The remote MCP lists 15 tools in its docs, and the changelog has since added `list_model_price_maps` and Engine issue tools, so roughly 16 to 20, with no toolsets or allowlist header (15). `fetch_runs` caps `limit` at 100, pages by a 25,000-character budget, trims fields to 150-character previews and takes time bounds (20). 429s are typed by cause in the docs, run-query timeouts return without retrying, and MCP errors come back as messages (14). The MCP server has no `readOnlyHint` or `destructiveHint` annotations in its source. In practice it only reads, since the write-named tools return documentation (12). Python and TypeScript SDKs. `LANGSMITH_TRACING=true` and a key are the only setup for LangChain apps (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, 81 out of 100, up to 3.8 more on the total

Why it scored 81: incident.io status page per region with 11 components on GCP US (20). For July to October the GCP US bars show 100 per cent for the API, application, PromptHub and LLM gateway, 99.97 per cent for run ingestion, 99.94 per cent for the deployments control plane and 99.83 per cent for Fleet. The individual incident list didn't render for us and we didn't check the EU, APAC or AWS pages, so this rests on the bars (20, minor only). Per-key limits published by endpoint (5,000 run writes, 30 `GET /runs/:id` and 2,000 other calls a minute) plus hourly event and data limits per plan (15). The docs explain each kind of 429 and recommend exponential backoff with jitter, and the SDKs retry and batch up to 100 runs per call. No `Retry-After` is documented (13). Enterprise has a support SLA. We found no uptime SLA (3). API and remote 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, 86 out of 100, up to 2.3 more on the total

Why it scored 86: Public OpenAPI 3.1 at api.smith.langchain.com/openapi.json with well over 100 paths, deprecated operations flagged (25). llms.txt and Markdown docs (10). MCP tool docstrings are long and practical. `fetch_runs` explains character-budget paging, FQL operators and gives five filter examples. But four tools named like actions (`push_prompt`, `create_dataset`, `update_examples`, `run_experiment`) only return how-to text, which the docstrings say and the names don't (16). Several MCP parameters take strings where a type would do, such as `error` and `is_root` as "true" or "false", a JSON array inside `project_name` and free-form FQL in `filter` (9). Examples throughout the docs and docstrings. The spec declares no 429 and error bodies get little documentation (11). Versioned `/api/v1` and `/v2` routes, a weekly dated changelog and a published deprecation policy (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. Transparency & trust, 78 out of 100, up to 1.9 more on the total

Made of editorial 70, provenance 86.

Why it scored 78: Closed platform. The SDKs and the open-source MCP server are MIT (18). Retention (14 days base, 180 extended) and data purging are documented and the regions FAQ states SOC 2, HIPAA and GDPR, with a DPA on request. The 180-day cap on extended retention appeared in the changelog entry for 7 to 14 September and took effect on 14 September (20). Published deprecation policy with a 6-month cloud window and `Deprecation` and `Sunset` headers, and the v1 runs endpoints carry a 31 January 2027 sunset (20). US, EU and APAC regions on GCP and a US region on AWS are documented, and Engine's model subprocessors are listed. We couldn't read a full subprocessor list (12).

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: langchain.com, registered 2019-12-03 (6 years) (11 of 15)
- security.txt: not found (0 of 10)

## 7. Maintenance & community, 85 out of 100, up to 1.3 more on the total

Why it scored 85: Python SDK v0.14.2 on 2026-09-30, with v0.14.3 and JS 0.10.7 following on 2 October (30). Twenty Python SDK tags since 30 July alone (20). The SDK repository takes commits daily and auto-labels new issues. We didn't load reply times (12). Python and TypeScript SDKs both current (15). CI, benchmarks, link checks and a guard on the generated OpenAPI client run in public. Pass state not checked (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.

## Deductions

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

- 2026-02-09 to 2026-06-19. Five published advisories in the LangSmith SDK. The most serious, GHSA-f4xh-w4cj-qxq8 (CVE-2026-59152, 19 June), let anyone who could send HTTP requests to a server running `TracingMiddleware` read arbitrary files from it into trace attachments, fixed in Python 0.8.18. GitHub's list rates it critical and the advisory scores it CVSS 5.0. The others cover untrusted public prompt manifests (high, 7 May), streaming tokens bypassing output redaction (moderate, 14 April), prototype pollution (moderate, 9 April) and SSRF through tracing headers (moderate, 9 February). All fixed and disclosed in public, so -4 (https://github.com/langchain-ai/langsmith-sdk/security)

## 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.

- The exact current MCP tool count. The docs list 15 and the changelog adds `list_model_price_maps` and Engine issue tools
- Incident details behind the status bars, and the EU, APAC and AWS status pages, which we didn't load
- unchecked: the subprocessor list and certifications in the trust centre, which didn't render for us
- Whether the `POST /feedback/eager` removal on 10 August 2026 had the six months' notice the policy promises. Its deprecation and removal appear in the same changelog entry

## Weaknesses

- Five SDK advisories from February to June 2026, one listed as critical (arbitrary file read through `TracingMiddleware`)
- Base traces keep 14 days, and extended retention was capped at 180 days with about a week's notice
- Closed-source platform, self-hosting, BYOC and key roles are Enterprise only
- Four MCP tools named like writes (`create_dataset`, `run_experiment` and others) only return documentation
- No uptime SLA found below Enterprise's support SLA

## 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.

- Batch run reads. `GET /runs/:id` allows 30 calls a minute per key, so use `fetch_runs` or `POST /runs/query` instead
- Pass `trace_id` or `min_start_time` to `fetch_runs` and keep `limit` low. Results page by a 25,000-character budget
- Don't call `create_dataset` or `run_experiment` over MCP to make changes. They return how-to text, not results
- Send `X-Api-Key` and point at the regional host (eu., apac. or aws.api.smith.langchain.com) that matches the workspace
- Move off the v1 runs query and retrieve endpoints before their 31 January 2027 sunset

## What the review panel asked for

- six months for retention
- rename the how-to tools
- declare 429 responses

## 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.