The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents how to run an audit script using the blender executable. This is standard for headless 3D processing and the script provided (scripts/audit_mesh.py) is a read-only data extraction tool that performs mesh integrity checks without making changes to the file system or executing untrusted code.- [SAFE]: The Python script scripts/audit_mesh.py strictly uses the bpy (Blender) API for analytical purposes. It validates geometry (non-finite coordinates, degenerate faces) and returns the data as JSON. It does not perform any network operations, sub-process spawning, or file writing.- [SAFE]: The logic in references/bpy-mesh-recipes.md demonstrates best practices for AI agents, such as avoiding global scene clears, using named datablocks to prevent collisions, and preserving world transforms during parenting. These practices reduce the risk of accidental data destruction during agent operations.- [SAFE]: The YAML frontmatter and configuration correctly scope the tool to mesh authoring and do not request excessive permissions or access to sensitive platform tools.

blender-mesh-authoring