viznoir is an AI-ready VTK visualization tool for the science and engineering domains, positioned to give AI agents full access to the VTK rendering pipeline, thereby enabling a workflow of "one prompt → physics analysis → cinematic rendering → LaTeX equations → publication-grade narrative."
Purpose viznoir mainly serves the visualization needs of simulation data such as CFD (computational fluid dynamics), FEA (finite element analysis), and SPH (smoothed particle hydrodynamics). It allows AI agents to read simulation data, apply filters, render high-quality images, and export animations in a headless environment, replacing the traditional approach that relies on the ParaView GUI or Jupyter Notebook.
Main capabilities
- MCP server: viznoir runs as an MCP (Model Context Protocol) server, providing AI agents with full access to the VTK rendering pipeline.
- Tools and resources: built-in 22 tools, 12 resources, and 4 prompts, covering the complete workflow from data reading, filtering, and rendering to animation export.
- Format compatibility: supports more than 50 file formats, including OpenFOAM, VTK, CGNS, Exodus, STL, glTF, and others.
- Physics-aware animation: provides seven presets that convert raw simulation data into publishable dynamic imagery, with each preset binding rendering primitives to a specific physical phenomenon.
- Story composition (compose_assets): can execute a workflow of inspect → render → annotate → compose → narrate, generating a four-panel physics breakdown figure from a single prompt, along with the governing equations rendered in LaTeX.
- Layout modes: supports story (vertical narrative), grid (N×M comparison), slides (16:9 keynote), and video (MP4 video with transitions).
- Native 3D annotation: annotations are rendered directly within the 3D scene via native VTK text actors and leader lines, without the need for Photoshop or matplotlib overlays.
Usage
- Use as an MCP client: add viznoir to the MCP client configuration (such as claude_desktop_config.json, ~/.cursor/mcp.json, etc.), then issue instructions to the AI agent,for example, "Open cavity.foam, render the pressure field with cinematic lighting, then create a physics breakdown story."
- Use as a Python library (advanced): all tool implementations can be imported as async functions. The user needs to provide a VTKRunner and await the results.
Applicable scenarios viznoir is suitable for scientific research and engineering scenarios that need to quickly turn simulation data into high-quality visualization results, and is especially suitable for users who want to use AI agents to automate the visualization workflow. It supports use with any MCP client such as Claude Code, Cursor, Windsurf, and Gemini CLI.
Limitations Using viznoir requires Python ≥3.10, the VTK wheel is installed automatically, and EGL headless rendering is supported. Regarding specific pricing models, prices, platform limitations, or performance guarantees, the materials do not currently provide this information.
