LangSmith MCP Server

Overview

The LangSmith MCP Server is a Model Context Protocol (MCP) server that integrates with LangSmith. It lets MCP-compatible clients (for example, AI coding assistants) read conversation history, prompts, runs and traces, datasets, experiments, and billing usage from your LangSmith workspace.

Tracing Capabilities

MCP Tool Call Tracing

Log every MCP tool call (with inputs and outputs) to a separate LangSmith project for monitoring and analytics. Each tool run is traced with:

• run_type="tool"
• session_id in metadata (from mcp-session-id, x-session-id, or x-request-id header)
• Generated per request for HTTP transport

Fetching Traces

Fetch LangSmith runs (traces, tools, chains, etc.) from one or more projects with support for:

• Filters (run_type, error, is_root)
• FQL (filter, trace_filter, tree_filter)
• Ordering options

Available Tools

• get_thread_history: Retrieve message history for conversations
• list_prompts: Fetch prompts from LangSmith with optional filtering by visibility (public/private) and limit
• get_prompt: Get specific prompts by name
• fetch_runs: Fetch LangSmith runs (traces, tools, chains, etc.) from one or more projects

Deployment Options

1. Hosted Version

Available at https://langsmith-mcp-server.onrender.com/mcp

• Authentication via LANGSMITH-API-KEY header

2. Local Installation

Can be installed via PyPI/uvx

3. HTTP Server

Deployable as an HTTP server using Docker, enabling remote access via the HTTP-streamable protocol

GitHub Repository

Available at github.com/langchain-ai/langsmith-mcp-server

Use Cases

• Monitoring MCP tool calls and agent behavior
• Debugging AI applications with comprehensive tracing
• Analyzing conversation patterns
• Dataset and experiment management
• Billing usage tracking
• Observability for production AI systems

Integration

Provides powerful observability integration for MCP-compatible AI applications with comprehensive support for monitoring and analyzing agent behavior.