langsmith-mcp-server

What it is and what it does

This package implements a Model Context Protocol (MCP) server that bridges language models to LangSmith, an observability platform for LLM applications. It exposes tools for querying conversation threads, managing prompts, fetching execution traces and runs, accessing datasets and experiments, and retrieving billing metrics—all through a standardized MCP interface that LLM clients can call. The server can run locally (via PyPI or from source) or be accessed as a hosted HTTP endpoint without local installation.

The package depends on fastmcp, langsmith, langchain-core, and uvicorn to handle MCP protocol transport, LangSmith API communication, and async HTTP serving. It requires Python 3.11 or later. While the repository is archived and no longer actively maintained, the package remains functional for integrating LangSmith data into MCP-compatible clients like Cursor or Claude Desktop.

Use it for:

• Query conversation history from LangSmith threads within an MCP client to provide context to language models.
• Retrieve and manage prompts stored in LangSmith workspaces programmatically through MCP tool calls.
• Fetch execution traces and runs from LangSmith projects to analyze LLM application behavior in real time.
• List and read datasets and examples from LangSmith for evaluation and testing workflows.
• Access experiment results and metrics (latency, cost, feedback) for dataset-based evaluations.
• Monitor organization billing usage and trace counts across workspaces via MCP queries.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Exposes LangSmith observability data (conversation history, prompts, runs, datasets, experiments, billing) through a Model Context Protocol server that language models can query.

Yes, if you actively use LangSmith and need MCP integration. The package is low-friction to install and permissively licensed. However, the archived repository and abandoned maintenance status mean no active bug fixes or feature development—use it for stable, read-heavy workflows (querying history, datasets, runs) rather than as a foundation for custom extensions. If you need a hosted solution, the Render instance is available but its long-term availability is not guaranteed.

Install

langsmith-mcp-server on PyPI

pip install langsmith-mcp-server

uv add langsmith-mcp-server

poetry add langsmith-mcp-server

Installing langsmith-mcp-server

Before you install

Low install friction with a pure-Python wheel and four runtime dependencies. However, the repository is archived and marked abandoned as of the latest commit, signaling no active maintenance despite recent release activity.

License in practice

MIT license permits commercial and private use with minimal restrictions, making it suitable for most deployment scenarios.

Quickstart

# Install from PyPI
pip install langsmith-mcp-server

# Add to MCP client config (e.g., Cursor mcp.json)
{
  "mcpServers": {
    "LangSmith MCP": {
      "command": "uvx",
      "args": ["langsmith-mcp-server"],
      "env": {
        "LANGSMITH_API_KEY": "your_api_key",
        "LANGSMITH_WORKSPACE_ID": "your_workspace_id"
      }
    }
  }
}

Requires Python 3.11 or later. LangSmith API key and workspace ID must be provided via environment variables. MCP client must support HTTP-streamable or subprocess-based server transport.

Verify before relying

• Whether the hosted instance at langsmith-mcp-server.onrender.com remains operational and reliable for production use.
• Current state of the archived repository and whether critical bugs or security issues are being addressed.

Package facts

Evidence: langsmith_mcp_server-0.1.1-py3-none-any.whl

Tags

More Artificial Intelligence packages