mcp-server-architect
by funtusov
The mcp-server-architect is a Model Context Protocol server that acts as an AI Software Architect. It analyzes codebases to generate Product Requirements Documents (PRDs) and provides reasoning assistance for complex coding tasks using a powerful agent-based architecture.
Last updated: N/A
What is mcp-server-architect?
The Architect MCP Server is a sophisticated agent-based system that mimics a human software architect. It leverages OpenAI's GPT-4o and other tools to analyze codebases, generate Product Requirements Documents (PRDs), and provide reasoning assistance for complex coding tasks.
How to use mcp-server-architect?
The server can be installed using uv or pip. After installation, it can be run directly or used with MCP Inspector for debugging. It can also be added to Claude Code in various scopes (local, project, user) to leverage its tools for PRD generation and reasoning assistance. API keys for OpenAI, Gemini, and Exa AI are required.
Key features of mcp-server-architect
Multi-Model Architecture (GPT-4o, Gemini)
Intelligent Codebase Analysis
Agent-Based Design with Autonomous Decision Making
Tool-Based Processing (Code Reader, Web Search, LLM Tool)
Comprehensive PRD Generation
Advanced Reasoning Assistance
MCP Integration with Claude Code
Logfire Instrumentation for monitoring
Use cases of mcp-server-architect
Generating Product Requirements Documents from existing codebases
Assisting developers with complex coding challenges by providing step-by-step reasoning
Analyzing codebases to understand architecture and dependencies
Integrating AI-powered architectural insights into development workflows
FAQ from mcp-server-architect
What API keys are required to run the server?
What API keys are required to run the server?
The server requires an OpenAI API key for GPT-4o, a Google API key for Gemini Pro, and an Exa API key for web search capabilities. A Logfire API key is optional for monitoring.
How do I install the server?
How do I install the server?
The recommended way to install the server is using uv. You can also install it using pip.
How do I provide the API keys to the server?
How do I provide the API keys to the server?
API keys can be provided as environment variables with the env prefix (e.g., env GEMINI_API_KEY=your_key mcp-server-architect) or through a .env file in the current directory.
How do I use the server with Claude Code?
How do I use the server with Claude Code?
You can add the server to Claude Code using the claude mcp add command, specifying the desired scope (local, project, or user) and providing the necessary API keys.
What tools does the server expose?
What tools does the server expose?
The server exposes two tools: Architect::generate_prd for generating PRDs and Architect::think for providing reasoning assistance.