higress-ops

Higress OPS MCP Server provides a Model Context Protocol (MCP) interface to configure and manage Higress instances. It includes a client built on LangGraph and LangChain MCP Adapters, enabling you to control Higress through an agent-driven workflow. The server exposes tools for common tasks such as managing routes, service sources, and request blocks, and provides a structured way to register new tools and integrate with the Higress Console API. You can interact with these tools via the MCP client, which leverages a standard tool registration pattern so you can extend capabilities by adding your own tool classes and methods.

To use the server, start the MCP client (which in turn starts the MCP server in stdio mode). Once running, you can invoke registered tools through the MCP client interface. Tools are organized by categories (e.g., CommonTools, RouteTools, ServiceSourceTools) and can be extended by adding new tool classes and methods that call Higress API endpoints. Some sensitive operations can be gated behind human confirmation by listing tool names in SENSITIVE_TOOLS. The README provides concrete guidance on adding new tools, integrating with the HigressClient, and registering them in server.py.

Prerequisites:

• Python 3.9+ installed
• Access to the Higress API (via Higress Console or equivalent)
• Git to clone the repository
• Optional: uv (via uvx) for running the MCP in stdio mode

Installation steps:

1. Clone the repository: git clone https://github.com/user/repo-higress-ops-mcp-server.git cd repo-higress-ops-mcp-server

2. Create and configure environment variables:

   • Copy the example env file if provided: cp .env.example .env
   • Edit .env to provide Higress API endpoint and credentials as needed.

3. Install Python dependencies: python -m pip install --upgrade pip pip install -r requirements.txt

4. Run the MCP client (which starts the MCP server in stdio mode): uv run client.py

Notes:

• The server is designed to be extended by adding new tool classes and registering them in server.py.
• If you prefer a different runtime (e.g., direct Python invocation), adapt the command in mcp_config accordingly.

Tips and common considerations:

• Ensure your .env file contains the necessary Higress API endpoint and credentials; the README mentions filling in environment values from .env.example.
• When adding new tools, place the class in the appropriate tools directory and register it in the tool_classes list in server.py.
• If a tool requires human confirmation, add its name to the SENSITIVE_TOOLS list to enforce an extra confirmation step before execution.
• The client/server architecture is designed to be extensible via the HigressClient wrapper methods (utils/higress_client.py). Implement API calls using the existing get/put/post helpers.
• Debug logging can help diagnose integration issues between the MCP server and the Higress API; ensure logger setup occurs as ToolsRegister initializes.

Environment variables to consider:

• HIGRESS_BASE_URL: Higress API base URL
• HIGRESS_API_TOKEN: Access token for API authentication
• Other application-specific variables your tools may require

If you encounter issues starting the server in stdio mode, verify that the path to client.py is correct and that uvx is installed and available in your environment.