Gemini Agent MCP Server

# Documentation for Commit b19a169 **Commit Hash:** b19a169a385e474a04c9b87a03d13a448d7ff8c0 **Commit Message:** feat: implement initial Gemini Agent MCP server **Generated:** Tue Oct 14 21:20:29 EDT 2025 **Repository:** geminiAgentMCP --- Here is the documentation for the provided git diff. *** ### **Subject: Initial Commit - Gemini Agent MCP Server (JSON-RPC)** #### **1. Summary** This commit introduces the "Gemini Agent MCP Server," a new Python-based service that exposes a suite of developer tools via a JSON-RPC 2.0 API. The server acts as a bridge between AI agents and common development tasks, leveraging the Gemini model for code and content generation, and integrating with external services like GitHub, linters, and package managers. This initial commit includes the complete project structure, dependencies, core server logic, and a comprehensive set of tools. #### **2. Changes** * **Project Scaffolding:** * **`LICENSE`**: The project is licensed under the MIT License. * **`README.md`**: Comprehensive documentation on setup, usage, and security has been created. * **`requirements.txt`**: Project dependencies (`Flask`, `pylint`, `requests`, `flask-jsonrpc`) are defined. * **Core Application (`src/main.py`):** * A Flask server is implemented to provide the API endpoints. * **`GET /mcp`**: An endpoint that serves the Model Context Protocol (MCP) manifest. This manifest describes all available tools, their purpose, and their expected inputs/outputs. * **`POST /jsonrpc`**: The primary endpoint for tool execution, adhering to the JSON-RPC 2.0 specification. It includes a web-browsable API for easier development and testing. * **Tool Implementation (`src/tools/`):** A modular toolset has been developed, with each module focusing on a specific domain: * **`call_gemini`**: A core function to interact with the Gemini model via its command-line interface. * **Code Analysis & Generation**: * `lint_code`: Lints Python (`pylint`) or JavaScript (`eslint`) files. * `generate_unit_tests`: Generates unit tests for a specified function or class. * `generate_docstrings`: Adds docstrings to Python code. * **Documentation**: * `summarize_docs`: Summarizes the content of one or more documentation files. * **Dependency Management**: * `analyze_dependencies`: Parses `requirements.txt` or `package.json` and fetches package summaries from PyPI or npm. * **GitHub Integration**: * `create_github_issue`: Creates a new issue in a specified repository. * `create_github_pr`: Creates a new pull request. * **Security (`src/tools/utils.py`):** * A `is_safe_path` utility function has been implemented to prevent directory traversal attacks. This check is enforced in all tools that accept file paths as input, ensuring that file operations are restricted to the project's directory. #### **3. Impact** This commit establishes a new, standalone service that provides a powerful and standardized interface for automating development workflows. AI agents and other automated systems can now perform complex tasks like code linting, test generation, and GitHub operations by making simple JSON-RPC calls. The MCP manifest allows for easy discovery of the server's capabilities. #### **4. Usage** **A. Setup** 1. **Install Dependencies:** ```bash pip install -r requirements.txt # For JavaScript linting, ensure eslint is globally installed # npm install -g eslint ``` 2. **Configure Environment Variables:** ```bash export GEMINI_API_KEY="YOUR_API_KEY" export GITHUB_TOKEN="YOUR_GITHUB_TOKEN" export MCP_PORT=5001 # Optional, defaults to 5001 ``` 3. **Run the Server:** ```bash python src/main.py ``` **B. Example: Calling a Tool** Use `curl` to send a JSON-RPC request to the `call_gemini` method. ```bash curl -X POST \ -H "Content-Type: application/json" \ -d '{ "jsonrpc": "2.0", "method": "call_gemini", "params": { "prompt": "What is the capital of France?" }, "id": 1 }' \ http://127.0.0.1:5001/jsonrpc ``` #### **5. Breaking Changes** Not applicable. This is the initial commit for a new project. #### **6. Migration Notes** Not applicable. This is the initial commit for a new project.