healkeiser/fxhoudinimcp

What it solves

It enables AI assistants to directly interact with SideFX Houdini, allowing them to build node networks, set up complex simulations, inspect USD stages, and manage rendering tasks. This removes the need for users to manually translate AI-generated suggestions into Houdini's interface by giving the AI direct access to the software's Python API.

How it works

The project implements a Model Context Protocol (MCP) server that acts as a bridge between an AI client (like Claude, Cursor, or VS Code) and a Houdini session. It uses Houdini's built-in hwebserver to execute commands on the main thread via hdefereval. The AI client communicates with the MCP server via stdio, which then relays the requests to Houdini over loopback HTTP.

Who it’s for

Technical artists, VFX artists, and 3D developers who use SideFX Houdini and want to leverage AI to automate scene setup, debug networks, or accelerate their creative workflow.

Highlights

• Extensive Toolset: Provides 215 tools covering graph intelligence, geometry (SOPs), simulation (DOPs), USD/LOPs, rendering, and VEX code validation.
• Broad Client Support: Compatible with any standard MCP client, including Claude, Gemini, Cursor, Windsurf, and VS Code.
• Integrated Documentation: Includes full-text search and page retrieval for the version-specific Houdini manual.
• Workflow Automation: Offers one-call setups for Pyro, RBD, FLIP, and Vellum simulations.
• Non-Destructive: Every tool call is registered as a single undo step in Houdini, preserving the user's current selection and viewport state.