deep-code-reasoning

This MCP server pairs Claude Code with Google's Gemini AI to provide a dual-model workflow for deep code reasoning. Claude handles local-context tasks, multi-file refactoring, and CLI-driven workflows, while Gemini brings a very large context window and execution capabilities to analyze long traces, logs, and distributed system behavior. The server orchestrates AI-to-AI conversations, escalation of complex analyses, and cross-model hypothesis testing, enabling you to inspect code paths, traces, and failures across services. To use it, configure the MCP server in Claude’s desktop integration or invoke the server directly via the node entrypoint; then leverage the provided tools to start conversations, escalate analyses, and trace execution paths across your codebase.

Prerequisites:

• Node.js 18 or later
• npm (comes with Node.js)
• Access to Gemini API (Gemini API key)

Step-by-step installation:

1. Clone the repository:

git clone https://github.com/Haasonsaas/deep-code-reasoning-mcp.git
cd deep-code-reasoning-mcp

2. Install dependencies:

npm install

3. Set up your Gemini API key:

cp .env.example .env
# Edit .env and add your GEMINI_API_KEY

4. Build the project (if applicable):

npm run build

5. Start the MCP server (adjust path to the built index.js if not using dist):

npm run start

6. Optional: configure Claude Desktop to connect to the MCP server using the provided command and path, and set GEMINI_API_KEY in the environment for the process.

Tips and notes:

• Ensure GEMINI_API_KEY is available in the environment where the MCP server runs.
• The MCP server is designed to integrate with Claude Desktop configuration; update the path in your Claude config to point to dist/index.js after build.
• If you change deployment paths, update the mcp_config command args accordingly.
• The server supports conversational, traditional, and cross-system analysis tools; refer to the README section on available tools for parameter formats.
• If you encounter connectivity or API quota issues with Gemini, verify your API key scope and ensure network access to Google services.
• For local development, you can simulate calls by using the example payloads shown in the tool documentation sections.