maya

The maya MCP server lets you manage and interact with multiple Autodesk Maya sessions from a single server. It discovers Maya sessions via Maya's command port, allowing you to run Python code in the sessions, create virtual Python modules for reusable functions, and stream stdout/stderr back to you. With multi-session support, you can target specific sessions or work with the default auto-selected session when only one is available. The server exposes tools that let you list active sessions, manually add sessions if discovery misses one, create code modules, and execute Python code with configurable result capture modes. This enables agent-driven workflows where you can orchestrate Maya tasks, run computations, and observe outputs in real time through the MCP resources.

Prerequisites:\n- Python installed on your system (recommended 3.8+).\n- Access to install Python packages (pip or uvx).\n- Autodesk Maya with a command port enabled for remote interaction (see Maya setup notes below).\n\nInstallation steps:\n1) Install the MCP server via uvx (recommended):\n\n uvx maya-mcp-server\n\n2) Alternatively, install from PyPI with pip (if uvx is not available):\n\n pip install maya-mcp-server\n\n3) Verify installation by running the server:\n\n uvx maya-mcp-server # or python -m maya_mcp_server (depending on package structure)\n\n4) Ensure Maya is configured to expose a Python command port. In Maya, run:\n\n import maya.cmds as cmds\n cmds.commandPort(name=":7002", sourceType="python")\n\n Or add the commandPort setup to userSetup.py for automatic startup.\n\n5) Connect Claude or your MCP client to the server using the provided mcp configuration.\n

Tips and notes:\n- The server supports multi-session discovery and does not require Maya-side modules; it uses Maya's command port to bootstrap interactions.\n- If discovery misses a session, use tools.add_session to manually register host:port details.\n- Tools available include: list_sessions (view active sessions and their session_keys), add_session (manual session registration), write_module (create a virtual Python module in Maya), and execute_code (run arbitrary Python code in a session with result capture options: NONE, JSON, RAW).\n- Resources for each session are accessible via maya://sessions/{session_key}/info and maya://sessions/{session_key}/output for session metadata and streaming output.\n- For best results, ensure your Maya session has a command port open on a known port and that network permissions allow connections from the MCP client.\n- If you need to run code across sessions, you can specify session_key with tools that support targeting.