by LangChain / LangSmith
Updated September 11, 2026
MCP100 verdict
Recommended hosted LangSmith integration for MCP clients that need traces, runs, datasets, experiments, prompts, and billing data through LangSmith's OAuth boundary.
MCP100 Assessment
Editorial verdict
Reviewed September 11, 2026
Task Fit
Mapped
Developer workflows
Trust
Reviewed
A dated editorial review is published on the MCP detail page.
Maintenance
Date unavailable
No dated repository or package update is currently stored.
Popularity
Signals unavailable
No independently comparable adoption metric is currently stored.
Evidence scope
No MCP-specific repository metric is stored
No MCP-specific package metric is stored
Stars observation date unavailable · Downloads observation date unavailable
Publisher activity observed September 11, 2026
Current source snapshot
Source relationship: First-party MCP
Reviewed September 11, 2026
Evidence-backed setup
Choose your MCP client. A client connects to the MCP server; the language model is selected inside that client. Only stored, source-backed configurations are shown.
Transport not recorded
No sufficient client-specific setup evidence has been stored from the publisher or client documentation.
Publisher-documented setup is not the same as MCP100 execution testing. “MCP100 logic checked” means the stored configuration shape was reviewed, not that third-party MCP software was run.
Evidence checked 2026-09-11T00:00:00.000Z
Type-appropriate, source-backed setup for this mcp server profile.
Loading optional publisher installation details…
No complete evidence-backed client setup steps are currently stored. Optional publisher documentation may provide additional context below.
Capability
Recommended hosted LangSmith integration for MCP clients that need traces, runs, datasets, experiments, prompts, and billing data through LangSmith's OAuth boundary.
Inspect LangSmith execution and tracing data.
Work with evaluation datasets, examples, and experiments.
Access prompt and other LangSmith workspace resources subject to the user's permissions.
Not measured yet
Estimated token footprint of the MCP tool schemas exposed to the agent.
Reviewed access
Review the exact access boundary before connecting this MCP to accounts, credentials, or sensitive data.
Related reviewed MCPs
Search, read, and receive recommendations from current public AWS documentation.
Choose this instead if
Choose AWS Documentation MCP instead if searching and reading current AWS service and API documentation.
Search, read, and receive recommendations from current public AWS documentation.
Retrieve current library and framework documentation for coding agents.
Choose this instead if
Choose Context7 MCP instead if current library and API documentation inside coding agents.
Retrieve current library and framework documentation for coding agents.
Discover and run MCP servers through Docker's MCP gateway.
Choose this instead if
Choose Docker MCP Gateway instead if running multiple MCP servers behind one stdio connection.
Discover and run MCP servers through Docker's MCP gateway.
LangChain / LangSmith is listed as First-party MCP in MCP100’s current evidence.
Current official resources:
Original product, capability, tool, and usage documentation from the MCP publisher. Installation sections appear under Install & Agents.
Title: LangSmith Remote MCP - Docs by LangChain
URL Source: https://docs.langchain.com/langsmith/langsmith-remote-mcp
Markdown Content:
The LangSmith Remote MCP is a Model Context Protocol (MCP) server hosted by LangSmith. It exposes the same tools as the standalone LangSmith MCP Server (conversation history, prompts, runs and traces, datasets, experiments, billing) without a separate deployment. Interactive MCP clients connect over OAuth with no API key or header configuration; programmatic clients can authenticate with a LangSmith API key via the X-Api-Key header.The Remote MCP is available on all LangSmith Cloud regions, on BYOC data planes, and on self-hosted LangSmith deployments running v0.16 or later (self-hosted additionally requires configuring a signing JWKS—see Self-hosted LangSmith). Self-hosted deployments on earlier versions should continue to use the standalone LangSmith MCP Server.
LangSmith Cloud:
| Region | URL |
|---|---|
| GCP US | https://api.smith.langchain.com/mcp |
| GCP EU | https://eu.api.smith.langchain.com/mcp |
| GCP APAC | https://apac.api.smith.langchain.com/mcp |
| AWS US | https://aws.api.smith.langchain.com/mcp |
The server discovers the rest of its OAuth metadata via RFC 8414 at /.well-known/oauth-authorization-server on the same host, so a compliant MCP client only needs the URL above.Self-hosted LangSmith:https://<your-langsmith-host>/api/mcp, where <your-langsmith-host> is the hostname of your LangSmith instance.BYOC:https://<data_plane_url>/api/mcp, where <data_plane_url> is the URL of your BYOC data plane.
The Remote MCP supports two authentication methods. Use OAuth for interactive MCP clients (Claude Code, Cursor, and similar), and an API key for programmatic or headless clients that can’t complete a browser-based login.
OAuth 2.1 with Dynamic Client Registration (RFC 7591) is the default for interactive clients. Compatible MCP clients register themselves automatically on first use—there is no client ID to provision and no API key to manage.After registration:
The session is scoped to your LangSmith user and workspace permissions—calls through the MCP server can only view what your account is permitted to view.
Send a LangSmith API key in the X-Api-Key header on every request. This suits backend services, scripts, and SDKs, for example, the AI SDK, where the interactive OAuth flow is not practical.Requests are authorized as the user that owns the API key, scoped to that key’s workspace and permissions—the same authorization the key has elsewhere in the LangSmith API. Tools that accept a workspace_id argument can target a specific workspace; otherwise the key’s own workspace is used.
The Remote MCP exposes the same tool surface as the standalone server:
get_thread_historylist_prompts, get_prompt_by_name, push_promptfetch_runs, list_projectslist_datasets, list_examples, read_dataset, read_example, create_dataset, update_exampleslist_experiments, run_experimentget_billing_usageSee the standalone server reference for parameter and pagination details—both servers share the same tool implementations.
If a client loses its session (for example, after revoking access in your LangSmith account, or if the refresh token is invalidated), trigger re-auth from the client:
/mcp, select langsmith, choose re-authenticate.Self-hosted LangSmith deployments on v0.16 or later expose the Remote MCP at https://<your-langsmith-host>/api/mcp. Once enabled, authentication and the tool surface are identical to LangSmith Cloud.
The Remote MCP and its OAuth Authorization Server are wired automatically when config.hostname is set, but they stay inert (404) until you provide a signing JWKS. This is the one piece of configuration LangSmith Cloud handles for you. To enable it:
step:config.signingJwks (stored in the chart secret), or as the key langsmith_signing_jwks in your existing secret:After upgrading, the OAuth discovery endpoints and /api/mcp become live. Verify with:
For deployments on earlier versions, run the standalone LangSmith MCP Server in your own environment and point its LANGSMITH_ENDPOINT at your self-hosted instance.
From the publisher README
Original client setup and configuration retained from the publisher documentation.
Publisher-documented clients
These names appear in the publisher's installation documentation. MCP100-verified configurations are shown separately above when available.
Add the server to your project’s .mcp.json (or run claude mcp add --transport http -s user langsmith https://api.smith.langchain.com/mcp to install it user-wide):
Then run /mcp and select langsmith to complete the OAuth flow. Tools become available as mcp__langsmith__<tool_name>.
dcode)Add the server to your user-level ~/.deepagents/.mcp.json file to make it available in every Deep Agents Code project, or add it to a project-level .mcp.json file for only that project. See the Deep Agents Code MCP tools docs for discovery locations and precedence rules.
Then complete the OAuth login flow in one of two ways:
/mcp, select langsmith, and follow the login prompt.Launch dcode, or restart an active session, to load the LangSmith MCP tools. In an interactive session, run /mcp to inspect server status and loaded tools.
Add to your Cursor mcp.json:
Cursor will prompt you to complete the OAuth flow on first use.
The LangSmith CLI authenticates against this same OAuth server, so langsmith auth login logs you in via the OAuth device flow—no API key required:
The CLI prints an activation URL; open it, approve, and the CLI completes login and stores the token per profile in ~/.langsmith/config.json. It then works against the same projects, traces, runs, datasets, experiments, and threads as the Remote MCP server.
For programmatic use from the AI SDK, authenticate with an API key via the X-Api-Key header and the built-in http (Streamable HTTP) transport:
Pass tools directly to streamText or generateText. The Remote MCP is stateless and responds with JSON over the standard Streamable HTTP transport, so the built-in transport works as-is—you don’t need a custom transport.
Any MCP client supporting the Streamable HTTP transport can connect with just the URL above—using OAuth 2.1 with dynamic client registration, or a LangSmith API key in the X-Api-Key header.