mcp

This MCP server provides a remote, streamable HTTP interface to Lambda Capture's macroeconomic data tools via the MCP protocol. It exposes tools such as a semantic search over macroeconomic data, and supports standard MCP features like listing available tools and invoking a specific tool with arguments. You can interact with it through the Remote MCP API endpoints, using calls to list_tools to discover available capabilities and tools/call to execute them. The example in the README demonstrates how to compose a JSON-RPC 2.0 request to call a tool named macroecon_semantic_search with a query_text and a max_results parameter, returning structured results suitable for downstream reasoning or display in a UI. Use the Authorization header with your access token to authenticate requests.

To use this server from your client (e.g., an OpenAI-style tool integration or Claude Desktop App), configure your MCP client to point at the remote MCP URL and provide your API key. You can also configure Claude to load this MCP as a remote tool via npx mcp-remote and pass the Authorization header in the configuration. The server is designed to be compatible with standard MCP flows and to work alongside the Lambda Capture semantic search tooling.

Prerequisites:

• An API key from Lambda Capture (for authenticating against the remote MCP server)
• Node.js 18+ (includes npx and npm) if you plan to install/run via Node, or Python 3.11+ if you plan to run a local Python MCP server
• Git (optional, for cloning the repository)

Node installation (remote MCP usage via Node):

1. Ensure Node.js is installed (includes npx and npm).
2. Install and run the remote MCP client via npx: npm install -g mcp-remote (optional if you want a global install) npx mcp-remote https://mcp.lambda-capture.com/v1/mcp/ --header "Authorization: Bearer YOUR_ACCESS_TOKEN"

Python/local server setup (if you have a local MCP server):

1. Create and activate a virtual environment: python -m venv .venv source .venv/bin/activate
2. Install dependencies: pip install -r requirements.txt
3. Run the local server (example, adjust to your entrypoint): python -m <your_mcp_module> # or python main.py depending on project layout

Configuration for clients (example in Claude Desktop App):

• Node-based remote usage: { "mcpServers": { "lambda-capture-mcp": { "command": "npx", "args": [ "mcp-remote", "https://mcp.lambda-capture.com/v1/mcp/", "--header", "Authorization: Bearer YOUR_ACCESS_TOKEN" ], "description": "RemoteMCP with Lambda Capture Macroeconomic Data API" } } }
• Python/local usage would mirror the local config shown in the README, with the appropriate command path to the Python interpreter and the path to your MCP entry script, including your API key in the environment or args as needed.

Tips and notes:

• Always supply your Authorization header with a valid Bearer token when calling the remote MCP endpoint.
• The server supports standard MCP operations such as list_tools and tools/call; use these to discover and invoke tools like macroecon_semantic_search.
• When configuring client integrations (e.g., Claude Desktop App), you can either run the remote MCP via npx or point the client to a locally hosted MCP server (Node or Python) by specifying the appropriate command, arguments, and environment variables.
• If you encounter authentication or CORS issues, verify that the token is valid and that the server URL matches the configured endpoint. Ensure your client includes the Accept header for JSON or JSON+Event-Stream responses as required by the MCP protocol.
• For local development, you may need to adjust max_tokens or context window settings on the client side to optimize performance with larger semantic search results.