seedance-anime-drama

Pass

Audited by Gen Agent Trust Hub on Sep 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local shell scripts (ofox-image.sh and ofox-video.sh) via bash to interact with the vendor's API for image and video generation. These scripts are located in sibling skill directories relative to the installation path.\n- [EXTERNAL_DOWNLOADS]: The skill relies on sibling skills (ofox-image-core and ofox-video-core) provided by the same author. It includes troubleshooting instructions that guide the user to install missing components using npx from the author's official GitHub repository.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied novel or script excerpts, which creates a surface for indirect prompt injection.\n
  • Ingestion points: Novel or script excerpts provided by the user as described in SKILL.md.\n
  • Boundary markers: Absent; the user text is interpolated into character and shot prompt templates without explicit delimiters or instructions to ignore embedded commands.\n
  • Capability inventory: Execution of shell scripts, network requests via curl (inside scripts), and file system writes to the specified output directory.\n
  • Sanitization: Absent; the skill does not define specific sanitization or validation for the user-supplied excerpts before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 22, 2026, 08:46 AM