skills/zju-real/easel/short-drama/Gen Agent Trust Hub

short-drama

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes subprocess.run to orchestrate the video production process. It executes local scripts such as ai_video.py and asr.py, and invokes standard system utilities like ffprobe. These operations are performed using list-based arguments, which is a standard method for internal tool coordination.\n- [INDIRECT_PROMPT_INJECTION]: The production pipeline processes user-supplied story ideas and synopses to generate full scripts and prompts for external AI generation services. While inherent to the skill's purpose, this represents a workflow that processes untrusted data.\n
  • Ingestion points: Theme and synopsis input in SKILL.md.\n
  • Boundary markers: None explicitly defined in the prompt generation logic.\n
  • Capability inventory: Subprocess execution and local file system access across scripts/drama_ops.py and scripts/dubbing.py.\n
  • Sanitization: Not explicitly implemented for user-provided text before script generation.\n- [EXTERNAL_DOWNLOADS]: The skill's documentation in EASEL-META.md references various external repositories, such as HBAI-Ltd/Toonflow-app and Forget-C/Jellyfish, for methodology. The skill does not perform any actual downloads or code execution from these remote sources at runtime.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 01:21 PM