short-drama-video-prompts

Pass

Audited by Gen Agent Trust Hub on Aug 19, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes several Python scripts (container_check.py, motion_timing_check.py, music_spec_check.py) intended for local execution to validate drama production data. These scripts perform structural, arithmetic, and sanity checks on JSONL project files and do not perform network operations or access sensitive system paths.\n- [PROMPT_INJECTION]: The skill processes untrusted external data in the form of JSONL files (storyboards, screenplay indexes, character assets), creating an indirect prompt injection surface.\n
  • Ingestion points: Data is ingested from files like shots.jsonl, motion-specs.jsonl, and short-drama.json located within the project structure.\n
  • Boundary markers: The skill employs a sources header system that uses SHA256 hashes to pin artifact versions and uses specific delimiters (fences) to separate verbatim dialogue from AI instructions.\n
  • Capability inventory: The agent generates prompts based on the provided data and utilizes the included validation scripts for ensuring consistency.\n
  • Sanitization: The provided Python scripts implement checks for duration arithmetic (VID-04) and container mapping (VID-15) to detect inconsistencies in the ingested data.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 19, 2026, 03:44 AM
Security Audit — agent-trust-hub — short-drama-video-prompts