mcp -js

This MCP server exposes YepCode tools and processes as configurable tools that AI agents can invoke through the Model Context Protocol. The server supports code execution and API management capabilities, and it can mix implementations in Python and Node.js while exposing each tool with its own parameter schema defined as JSON Schema. The base example shows how to run the server via NPX; you can replace the NPX invocation with Docker or other deployment hooks as needed. Tools such as run_code are available to execute user-provided code securely in YepCode’s environment, and additional API management tools (yc_api and yc_api_full) allow controlling processes, schedules, variables, storage, and more. When enabled, process-tagged tools mirror your YepCode processes, exposing them as individual MCP tools that you can call with structured inputs to drive automation workflows.

Prerequisites:\n- Node.js (version 18 or higher) or access to NPX via npm.\n- A YepCode API token (get from YepCode Cloud under Settings > API credentials).\n\nInstall and run using NPX:\n1. Ensure Node.js is installed.\n2. Use NPX to run the MCP server configuration shown in the example:\n\nbash\n# Example: NPX-based run (from README)\n``\n3. Provide your YepCode API token in the environment variable when running or via the config file as shown in the configuration example.\n\nAlternative: Docker deployment:\n1. Build the container image:\nbash\ndocker build -t yepcode/mcp-server .\n\n2. Run with a configuration similar to the following in your environment:\ntypescript\n{\n "mcpServers": {\n "yepcode-mcp-server": {\n "command": "docker",\n "args": [\n "run",\n "-d",\n "-e",\n "YEPCODE_API_TOKEN=your_api_token_here",\n "yepcode/mcp-server"\n ]\n }\n }\n}\n```\n\nPrerequisites recap: ensure you have the API token and either NPX access or a running Docker environment to deploy the server.

Tips and common considerations:\n- Keep YEPCODE_API_TOKEN secure; avoid committing it to source control. Consider using a secrets manager or environment file.\n- The server can be extended to expose specific processes by tagging them in YepCode and configuring YEPCODE_MCP_TOOLS via environment variables to limit which tools are available.\n- If debugging, use the MCP Inspector to observe stdio communication and inspect tool inputs/outputs.\n- If you enable run_code options, you can control behavior with YEPCODE_MCP_OPTIONS (e.g., runCodeCleanup, skipCodingRules) to tailor tool definitions and behavior.\n- The hosted cloud URL is https://cloud.yepcode.io/mcp for a hosted MCP server experience without local installation.