The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill acts as a code generator that transforms user requests into executable Blender Python scripts. This design creates a surface for indirect prompt injection where malicious instructions could be embedded in the data being processed.
Ingestion points: User queries (e.g., "animate this", "make it move") and scene data retrieved via mcp__blender__get_scene_info or mcp__blender__get_object_info in SKILL.md.
Capability inventory: The skill utilizes mcp__blender__execute_blender_code to run generated Python scripts and Bash for command execution.
Boundary markers: The instructions do not define clear boundary markers or sanitization logic to prevent user-controlled strings (like object names) from escaping their intended context in the generated code.
[DYNAMIC_EXECUTION]: The skill utilizes and documents Blender's "Scripted Drivers" functionality, which involves the runtime evaluation of Python expressions to drive animation properties.
Evidence: Recipe 6 in SKILL.md demonstrates setting driver.type = 'SCRIPTED' and assigning a Python expression string to driver.expression.
Evidence: The references/overview.md file further details the use of 'SCRIPTED' drivers as a professional animation technique, which involves executing arbitrary Python expressions every frame.

blender-animation