seedance-short-drama
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided scripts and dialogue to generate complex prompts for a video generation model.\n
- Ingestion points: Scene descriptions and dialogue beats are ingested through a creative brief process and script analysis in SKILL.md.\n
- Boundary markers: The skill employs a structured manifest template and uses quotation marks for dialogue sections to define boundaries for the AI model.\n
- Capability inventory: The skill possesses the ability to execute a local shell script (ofox-video.sh) and perform network operations via curl to the Ofox API infrastructure.\n
- Sanitization: The skill lacks explicit programmatic sanitization of user input, relying instead on detailed prompt instructions and negative lists (AVOID) to guide model output.\n- [COMMAND_EXECUTION]: The skill executes a vendor-provided local script, ofox-video.sh, to perform environment checks, calculate costs, and manage the video generation lifecycle. The script is invoked using relative and absolute paths within standard skill installation directories.\n- [EXTERNAL_DOWNLOADS]: The skill interacts with the Ofox API (api.ofox.ai) to submit generation jobs and download resulting media. These network operations are directed at the official domains of the skill's authoring organization.
Audit Metadata