The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/run_blender.py launches the Blender executable using subprocess.Popen. This is the core functionality of the skill and is implemented following security best practices. The command includes flags like --background for headless operation, --factory-startup to ignore local user configurations, and --disable-autoexec to block the execution of potentially malicious Python scripts or drivers embedded in user-supplied .blend files.
[SAFE]: The skill includes a robust validation mechanism in scripts/evidence.py that uses the ast module to perform static analysis on any Python scripts generated by the agent. This mechanism ensures that the code only uses supported Blender APIs and is strictly bound to specific 'evidence' passages read from the bundled API corpus, providing a high level of traceability and preventing the execution of unverified or hallucinated commands.

blender-cli