Repository navigation
3D Visualizer for Python and MCP 0.4.2
3D Visualizer is now available on PyPI as 3d-visualizer. The Python import remains ply_visualizer; existing ply-viewer* command aliases are retained.
Install the Python library with uv add 3d-visualizer. Start MCP with:
uvx --from '3d-visualizer[mcp]==0.4.2' 3d-visualizer-mcp --root /absolute/path/to/dataThe MCP registry identifier is io.github.kleinicke/3d-visualizer. This release adds registry metadata, public command aliases and an MCPB 0.4 UV bundle. The attached bundle installs the pinned PyPI package and asks for the allowed data directory. Inline interactive rendering requires an MCP Apps host with WebGL; a browser fallback is available.
Validation: 40 Python tests completed with one CUDA-dependent skip, strict wheel/sdist checks, a real stdio initialization discovering 27 tools from a clean installation, and MCPB manifest validation.