The Agent Skills Directory

[SAFE]: The skill implements best practices for 3D scene management, including a reset_world function to prevent state corruption and specific code paths to maintain compatibility across Blender versions 4.x and 5.x.
[COMMAND_EXECUTION]: The orchestrator generates and executes Python code via the mcp__blender__execute_blender_code tool. This is the core functionality required to drive the Blender application and is accompanied by detailed instructions on object naming and resource management.
[EXTERNAL_DOWNLOADS]: Guidance is provided for fetching 3D models from established industry services like Poly Haven and Sketchfab, which is a routine and safe practice for professional asset creation.
[PROMPT_INJECTION]: The skill processes untrusted natural language requests to generate code, creating an attack surface for indirect prompt injection. Ingestion points: User requests for 3D model creation or scene modification. Boundary markers: Not explicitly defined for input interpolation. Capability inventory: Includes mcp__blender__execute_blender_code and Bash for file system verification. Sanitization: The skill mandates a visual validation checkpoint using mcp__blender__get_viewport_screenshot to confirm the generated output matches the user's intent before completion.

text-to-blender