LangSmith

v1.0.0Monitoring & Observabilitystable

Enables language models to access LangSmith observability platform features including fetching conversation history, managing prompts, retrieving traces and runs, working with datasets and examples, and analyzing experiments.

langsmith-mcp-servermcpai-integration
Share:
113
Stars
0
Downloads
0
Weekly
0/5

What is LangSmith?

LangSmith is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to enables language models to access langsmith observability platform features including fetching conversation history, managing prompts, retrieving traces and runs, working with datasets and examples, a...

Enables language models to access LangSmith observability platform features including fetching conversation history, managing prompts, retrieving traces and runs, working with datasets and examples, and analyzing experiments.

This server falls under the Monitoring & Observability category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • Enables language models to access LangSmith observability pl

Use Cases

Access LangSmith observability platform features from Claude.
Retrieve conversation history, prompts, traces, runs, and experiments.
Manage datasets and analyze LLM application performance and behavior.
langchain-ai

Maintainer

LicenseMIT License
Languagepython
Versionv1.0.0
UpdatedMay 22, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx langsmith-mcp-server

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use LangSmith

The LangSmith MCP Server bridges AI assistants and the LangSmith observability platform, letting language models query traces, runs, prompts, datasets, and experiments directly from a conversation. Instead of switching between the LangSmith web UI and your AI client, you can ask questions about your LLM application's behavior, retrieve specific runs for debugging, compare experiment results, and manage prompt templates — all through natural language. It is ideal for teams monitoring production LLM applications or iterating on prompt engineering workflows.

Prerequisites

  • Python 3.9 or later with uv package manager installed
  • A LangSmith account with an API key (sign up at smith.langchain.com)
  • An MCP-compatible client such as Claude Desktop or Claude Code
  • Optional: a self-hosted LangSmith instance URL if not using the cloud service
1

Install uv package manager

The LangSmith MCP server uses uv for dependency management. Install it with the official installer script if you don't have it already.

curl -LsSf https://astral.sh/uv/install.sh | sh
2

Install the LangSmith MCP server package

Install the server using uv. This pulls the langsmith-mcp-server package and all its dependencies into an isolated environment.

uv pip install --upgrade langsmith-mcp-server
3

Obtain your LangSmith API key

Log in to smith.langchain.com, go to Settings > API Keys, and create a new key. Optionally note your workspace ID if you want to scope the key to a specific workspace.

4

Configure Claude Desktop

Add the server to your Claude Desktop configuration. Set LANGSMITH_API_KEY in the env block. Use LANGSMITH_WORKSPACE_ID to restrict to a specific workspace, or LANGSMITH_ENDPOINT for a self-hosted instance.

{
  "mcpServers": {
    "langsmith": {
      "command": "uvx",
      "args": ["langsmith-mcp-server"],
      "env": {
        "LANGSMITH_API_KEY": "your-langsmith-api-key",
        "LANGSMITH_WORKSPACE_ID": "optional-workspace-id"
      }
    }
  }
}
5

Restart Claude Desktop

Restart Claude Desktop to load the new MCP server. Once connected, you can query LangSmith data directly from the Claude chat interface.

6

Alternatively, use the hosted endpoint

A pre-deployed instance is available if you prefer not to run the server locally. Configure your MCP client to connect via the hosted URL using HTTP headers for authentication.

{
  "mcpServers": {
    "langsmith": {
      "url": "https://langsmith-mcp-server.onrender.com/mcp",
      "headers": {
        "langsmith-api-key": "your-langsmith-api-key"
      }
    }
  }
}

LangSmith Examples

Client configuration

Claude Desktop config using uvx to run the LangSmith MCP server with API key authentication.

{
  "mcpServers": {
    "langsmith": {
      "command": "uvx",
      "args": ["langsmith-mcp-server"],
      "env": {
        "LANGSMITH_API_KEY": "your-langsmith-api-key",
        "LANGSMITH_WORKSPACE_ID": "optional-workspace-id",
        "LANGSMITH_ENDPOINT": "https://api.smith.langchain.com"
      }
    }
  }
}

Prompts to try

Example natural-language queries that leverage the LangSmith MCP server's tools.

- "Fetch the last 10 root runs from my 'production-chatbot' project and summarize any errors"
- "List all public prompts in my workspace"
- "Get conversation history from thread 'thread-abc123' in project 'customer-support'"
- "List all datasets of type 'chat' and show me the example count for each"
- "Show me the billing usage for my organization for May 2025"
- "Retrieve the latest experiments from project 'gpt4-eval' and compare their pass rates"

Troubleshooting LangSmith

Authentication error: LANGSMITH_API_KEY not accepted

Verify the key is active in the LangSmith UI under Settings > API Keys. For the hosted endpoint, the header name must be lowercase 'langsmith-api-key' (not 'Authorization: Bearer ...'). For STDIO mode the env var must be LANGSMITH_API_KEY.

Empty results when querying runs or traces

Check that your API key has access to the correct workspace. Set LANGSMITH_WORKSPACE_ID to the specific workspace UUID where your traces are stored. If using a self-hosted instance, also set LANGSMITH_ENDPOINT to your instance URL.

uvx command not found when starting the server

uv ships with uvx as a companion command. Re-install uv with: curl -LsSf https://astral.sh/uv/install.sh | sh and then open a new terminal shell so the PATH is updated. Verify with: uvx --version.

Frequently Asked Questions about LangSmith

What is LangSmith?

LangSmith is a Model Context Protocol (MCP) server that enables language models to access langsmith observability platform features including fetching conversation history, managing prompts, retrieving traces and runs, working with datasets and examples, and analyzing experiments. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install LangSmith?

Follow the installation instructions on the LangSmith GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with LangSmith?

LangSmith works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is LangSmith free to use?

Yes, LangSmith is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.

Browse More Monitoring & Observability MCP Servers

Explore all monitoring & observability servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "langsmith-mcp-server": { "command": "npx", "args": ["-y", "langsmith-mcp-server"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use LangSmith?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides