The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/validate_creative_artifacts.py executes a shared validation tool via subprocess.run. The command uses a fixed path to an internal script and handles arguments safely without shell expansion.
[PROMPT_INJECTION]: The skill transforms user input into cinematic plans. It includes safety guidelines to prevent generating harmful content. While the input surface exists, no malicious patterns were observed.
Ingestion points: user_request in src/index.js.
Boundary markers: Not used in string interpolation.
Capability inventory: subprocess.run (validation scripts), file system access (shared utilities).
Sanitization: safeText utility from the shared runtime.

film-director