GemForge

GemForge is an MCP server that provides enterprise-grade Gemini integration for MCP agents like Claude, Roo Code, Windsurf, and related tools. It exposes a set of specialized tools (gemini_search, gemini_reason, gemini_code, gemini_fileops) that enable real-time web access, multi-file analysis, step-by-step reasoning, and code/document processing. The server automatically selects appropriate Gemini models for each task and includes robust error handling and retry logic to maintain resilience under rate limits or transient failures. To use it, deploy the server (for example via the provided Node.js setup) and authenticate requests with your Gemini API key. Then invoke the available tools through the MCP interface, passing appropriate parameters such as search queries, code analysis directives, or file paths for comparison. The system is designed to handle large repositories and mixed file types, making it suitable for enterprise-scale analysis and integration workflows.

Prerequisites:

• Node.js and npm installed on the host (recommended latest LTS).
• Access to the internet for installing dependencies from npm registry.

Installation steps:

1. Install dependencies and set up the MCP server:

# Option A: One-line setup with npx (recommended for quick start)
npx @gemforge/mcp-server@latest init

2. If you prefer manual setup, install the package and start the server:

# Install the GemForge MCP package
npm install gemforge-mcp

# Start the MCP server
npm start

3. Create the configuration file for GemForge (example shown in the project):

{
  "mcpServers": {
    "GemForge": {
      "command": "node",
      "args": ["./dist/index.js"],
      "env": {
        "GEMINI_API_KEY": "your_api_key_here"
      }
    }
  }
}

4. Validate that the server is running and accessible via MCP tooling or your MCP dashboard.

Notes:

• If you deploy via Docker or other environments, adapt the command and environment variables accordingly.
• Ensure GEMINI_API_KEY has the appropriate permissions for the Gemini models you intend to use.

Environment variables and configuration tips:

• GEMINI_API_KEY: Required for authenticating with Gemini services.
• GEMINI_PAID_TIER: Optional; set to true if you’re using paid tier for better rate limits.
• DEFAULT_MODEL_ID: Optional default model selection (e.g., gemini-2.5-pro).
• LOG_LEVEL: Optional logging level (debug, info, warn, error).

Model and tool behavior:

• gemini_search is optimized for real-time data retrieval and web access.
• gemini_reason provides step-by-step reasoning for complex problems.
• gemini_code performs deep code understanding and repository-level analysis.
• gemini_fileops handles 60+ file formats for multi-file processing and comparison.

Common issues:

• API key missing or invalid: ensure GEMINI_API_KEY is set and valid.
• Network or rate-limit errors: rely on the built-in retry/backoff logic; consider increasing LOG_LEVEL to debug for troubleshooting.
• Model selection: if default models don’t meet performance needs, override with model_id in tool calls or set DEFAULT_MODEL_ID in the environment.