The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/render_handoff.py executes Blender rendering operations using the bpy (Blender API). This is the primary and intended function of the skill.
[DATA_EXPOSURE]: The skill generates local manifest files (seam_sequence.json and handoff.json) which contain metadata about the rendered shots (paths, camera settings, frame ranges). These are stored in a user-defined output directory and do not contain sensitive system information.
[DYNAMIC_EXECUTION]: The skill uses runpy.run_path in its documentation examples to demonstrate how to execute its internal rendering script. This is a standard method for modular execution of local skill scripts.
[INDIRECT_PROMPT_INJECTION]: The script ingests user-provided arguments (paths, object names, and descriptions). It includes a sanitization helper _safe_id that uses regular expressions to clean folder names, mitigating common path injection risks.
[SAFE_PRACTICE]: The Python script implements a robust state management class _State that captures Blender's scene configuration (engine, resolution, camera, etc.) and restores it in a finally block, ensuring the user's environment is not permanently altered by the render process.

ai-video-handoff