apic

The apic-mcp-server exposes IBM API Connect capabilities to MCP clients and AI agent workflows. It provides a collection of tooling endpoints that mirror API Connect analytics and management features, enabling MCP clients such as Claude Desktop, VS Code, and IBM BOB to connect, configure, and run API Connect operations within their MCP workflows. With this server, you can access analytics tooling, management tasks, and other APIC capabilities via MCP JSON configurations and the client side mcp.json templates described in the repository. The server is designed to plug into your MCP client environment so you can wire in your APIC instance details, client credentials, and connection endpoints to drive automated API Connect tasks from conversations or automation pipelines.

To use it, deploy the MCP server in your preferred environment (via npm/pnpm or your MCP orchestration), then create an MCP server entry for apic-mcp-server in your agent's MCP config. The available tools section documents the Analytics and Management capabilities you can invoke. Configure environment variables as needed (for example, LOG_LEVEL to capture debug logs) and point the server to your APIC provider’s endpoint, API key, client ID, and client secret. Once running, you can invoke the server’s tools from your MCP clients by sending the appropriate mcp.json payloads and using the pre-defined tool links and templates included in the repository.

Prerequisites:\n- Node.js and npm installed on your system (LTS version recommended).\n- Access to an IBM API Connect environment (v10 or v12) with provider organization, API key, client ID/secret, and APIC endpoints.\n\nInstallation steps:\n1) Install Node.js and npm if you do not have them installed.\nbash\n# On macOS / Linux via a package manager (example with nvm and Node.js install):\nnvm install 0.38.0 2>/dev/null || true\nnode -v\nnpm -v\n\n2) Install the MCP server package from npm (or use npx as described in mcp_config).\nbash\n# Global install is not required if using npx in configuration; this is an alternative:\nnpm install -g apic-mcp-server # if you prefer a global install\n\n3) Run the MCP server via the MCP config (npx is used in the configuration):\nbash\nnpx -y apic-mcp-server\n\n4) If you are packaging for Claude Desktop or other MCP clients, ensure you place the generated .mcpb or .mcp.json template in the expected service folder (e.g., analytics) and configure the client with your APIC details as shown in the repository examples.\n\nNotes:\n- If you are using a tarball package or a .mcpb installer, you can point the MCP client to the local installer path or package accordingly.\n- You may need to set environment variables such as LOG_LEVEL for debugging, as described in the logs section of the README.

Tips and common considerations:\n- Logs: INFO level logs are written to a daily rotating file under your home directory (default: apic-mcp/apic-mcp-YYYY-MM-DD.log). For debug logging, set LOG_LEVEL=debug in your MCP client environment configuration.\n- When using NPM/NPX, ensure network access to fetch the apic-mcp-server package or tarball.\n- You can customize the MCP server entry name inside the mcp_config to better reflect your environment (e.g., apic-analytics-mcp-server).\n- Ensure your APIC provider details (provider organization, API key, client id/secret, and endpoints) are kept secure and not committed to version control.\n- Review the Suggested MCP Clients section in the README to pick the best integration flow for VS Code, Claude Desktop, or IBM BOB, and follow the setup steps for the chosen client.