What features does Lambda Capture MCP server provide?

Q: What features does Lambda Capture MCP server provide?

Remote MCP Server (streamable HTTP): A remote, streamable MCP endpoint at https://mcp.lambda-capture.com/v1/mcp/ with status checks, enabling JSON-RPC tool calls such as tools/call and list_tools.. OpenAI Responses API integration: Demonstrates how to use the OpenAI Responses API to call MCP as a tool (type: mcp) within a response, enabling macroeconomic semantic search workflows.. JSON-RPC endpoints (tools/call and list_tools): Supports JSON-RPC 2.0 methods to invoke semantic search tools and query available tools, accessed via the /v1/mcp/ endpoint.. Claude Desktop App integration: Guides configuring Claude Desktop App to connect to the remote or local MCP server via claude_desktop_config.json, including Node and Python backend configurations.. Local MCP Server - Node setup: Node-based local deployment with prerequisites and steps: npm install and npm run build to run the MCP server locally.. Local MCP Server - Python setup: Python-based local deployment with a virtual environment and dependencies installation (virtualenv, pip install -r requirements.txt).. Context window tuning: Guidance to adjust context size by changing maxTokens (TS) or max_tokens (Python) to fit the model's context window.. Security and API keys: Access is protected via Authorization: Bearer YOUR_ACCESS_TOKEN headers and a Lambda Capture API key.

Lambda Capture MCP Server 2025: Features & Setup Guide

Overview

Lambda Capture MCP Server is a remote and local implementation of the Macroeconomic Semantic Search API. The remote server at mcp.lambda-capture.com provides a streamable HTTP interface for performing semantic searches over macroeconomic data and returns results via JSON-RPC calls (e.g., tools/call with a macroecon_semantic_search tool and list_tools). The server is demonstrated for integration with the OpenAI Responses API, showing how to invoke the MCP as a tool within a response. Examples include POST requests to /v1/mcp/ using jsonrpc 2.0 payloads and curl to call the search tool or list available tools. The README also explains configuring clients such as Claude Desktop App to connect to the remote server or a local install via claude_desktop_config.json (Node and Python backends). For local development, prerequisites cover Node.js 18+ or Python 3.11+ with installation steps for Node (npm install, npm run build) and Python (virtualenv, pip install -r requirements.txt). It also notes adjusting context window size via maxTokens/max_tokens and securing access with Bearer tokens and a Lambda Capture API key.

Features

Remote MCP Server (streamable HTTP)

A remote, streamable MCP endpoint at https://mcp.lambda-capture.com/v1/mcp/ with status checks, enabling JSON-RPC tool calls such as tools/call and list_tools.

OpenAI Responses API integration

Demonstrates how to use the OpenAI Responses API to call MCP as a tool (type: mcp) within a response, enabling macroeconomic semantic search workflows.

JSON-RPC endpoints (tools/call and list_tools)

Supports JSON-RPC 2.0 methods to invoke semantic search tools and query available tools, accessed via the /v1/mcp/ endpoint.

Claude Desktop App integration

Guides configuring Claude Desktop App to connect to the remote or local MCP server via claude_desktop_config.json, including Node and Python backend configurations.

Local MCP Server - Node setup

Node-based local deployment with prerequisites and steps: npm install and npm run build to run the MCP server locally.

Local MCP Server - Python setup

Python-based local deployment with a virtual environment and dependencies installation (virtualenv, pip install -r requirements.txt).

Context window tuning

Guidance to adjust context size by changing maxTokens (TS) or max_tokens (Python) to fit the model's context window.

Security and API keys

Access is protected via Authorization: Bearer YOUR_ACCESS_TOKEN headers and a Lambda Capture API key.