Viznoir

What kimimgo/viznoir MCP server does

The kimimgo/viznoir MCP server exposes VTK visualization workflows to MCP clients. Agents can inspect scientific datasets, examine fields and timesteps, apply data filters, render scenes, probe values, compare outputs, and create animations. The repository lists 22 tools, 12 resources, and four prompts.

Its supported workflow covers both analysis and presentation. Analysis tools include data inspection, physics inspection, statistics extraction, line plotting, surface integration, and time-series probing. Visualization tools include standard and cinematic rendering, slicing, contouring, clipping, streamlines, isosurfaces, and volume rendering. Outputs can include PNG, WebP, MP4, glTF, and LaTeX-based assets.

How it works

The server connects an MCP request to a VTK pipeline containing readers, filters, rendering, and camera controls. Rendering is designed for headless execution using EGL or OSMesa, so the workflow does not depend on a GUI, ParaView, Jupyter notebook, or display server.

The repository describes physics-oriented animation presets for streamline growth, clip-plane sweeps, layer reveals, isovalue sweeps, structural warp oscillation, light orbits, and threshold reveals. The compose_assets capability can arrange generated assets as a story, grid, slides, or video composition. A Python library interface is also available: tool implementations can be imported as asynchronous functions and used with a VTKRunner.

Setup and configuration

Install the Python package with:

pip install viznoir

Python 3.10 or newer is required. Optional extras are available for mesh handling, composite animation, or the complete set of extras:

pip install "viznoir[mesh]"
pip install "viznoir[composite]"
pip install "viznoir[all]"

The MCP entry point is mcp-server-viznoir. A client configuration can set VIZNOIR_DATA_DIR to the simulation-data directory and VIZNOIR_OUTPUT_DIR to the output directory. The README shows this command-based configuration for Claude Desktop and Cursor, and says the server works with Claude Code, Cursor, Windsurf, Gemini CLI, and other MCP clients.

Tools and capabilities

The kimimgo/viznoir MCP server groups its tools into these areas:

• Rendering: standard, cinematic, batch, and volume rendering
• Filters: slicing, contouring, clipping, streamlines, and isosurfaces
• Analysis: dataset inspection, physics inspection, statistics, and analysis
• Probing: line plots, surface integration, and time-series probes
• Animation: animation and split animation
• Comparison: asset comparison and composition
• Export: 3D previews and pipeline execution

The README reports support for six VTK data types, 19 native file formats, and more than 50 formats through meshio. The mesh extra installs meshio and trimesh; the composite extra installs Pillow and matplotlib.

Limitations and notes

The server is focused on scientific visualization through VTK rather than general-purpose file or database operations. Some format coverage depends on optional extras, and the repository distinguishes native VTK formats from formats supplied through meshio. Rendering and animation also depend on the local Python, VTK, and headless graphics environment. The project is released under the MIT license.