mcp

MCP Server

The Model Context Protocol (MCP) server exposes LayerLens capabilities as tools that can be consumed by Claude, Cursor, VS Code Copilot, and any other MCP-compatible client. This enables AI assistants to directly query traces, run evaluations, and manage judges through natural language -- turning LayerLens into an interactive quality assurance co-pilot within your development environment.

Prerequisites

pip install layerlens --index-url https://sdk.layerlens.ai/package mcp
export LAYERLENS_STRATIX_API_KEY=your-api-key

Quick Start

Run the server in stdio mode for use with Claude Code or other MCP clients:

python layerlens_server.py

The server will start and wait for MCP protocol messages on stdin/stdout.

Available Tools

Tool	Description
list_traces	List recent traces with optional filters (date range, status, model).
get_trace	Retrieve a single trace by ID, including all spans and metadata.
run_evaluation	Run a trace evaluation using a specified judge.
get_evaluation	Fetch evaluation results by evaluation ID.
create_judge	Create a new AI judge with custom criteria.
list_judges	List all available judges in the workspace.

Configuration

Claude Code

Add the following to your MCP configuration file:

{
  "mcpServers": {
    "layerlens": {
      "command": "python",
      "args": ["samples/mcp/layerlens_server.py"],
      "env": {
        "LAYERLENS_STRATIX_API_KEY": "your-api-key"
      }
    }
  }
}

Other MCP Clients

Any MCP-compatible client can connect to this server using stdio transport. Consult your client's documentation for the configuration format, and point it to layerlens_server.py with the required environment variable.

Expected Behavior

Once connected, the MCP client will discover the available tools and make them accessible through its interface. For example, in Claude Code you can ask "list my recent traces" or "evaluate trace abc-123 with the safety judge" and the assistant will invoke the corresponding LayerLens tool.