The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill uses runpy.run_path to load and execute Python scripts from other local skills (stylized-style, lighting, rendering, etc.). The paths are constructed dynamically using string interpolation of a skills directory variable.
[INDIRECT_PROMPT_INJECTION]: The skill accepts a target identifier for a Blender object which is interpolated directly into script arguments. This creates a surface where unverified object names could influence the execution of downstream scripts.
Ingestion points: The target variable within the Python script block in SKILL.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded content are used for the interpolated target.
Capability inventory: File execution via runpy.run_path, scene object modification, and render parameter control.
Sanitization: No sanitization or validation logic is applied to the target string before it is passed to dependencies.

vector-style