novel-to-video
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
ffmpeg video skillto perform the final assembly of video clips, which involves executing system-level command-line tools to process and concatenate media files. - [INDIRECT_PROMPT_INJECTION]: The workflow relies on reading and processing user-supplied novels and scripts via the
novel-readerskill, creating a surface for indirect prompt injection. - Ingestion points: The
novel-readerskill ingests external literary text files which are then used to generate storyboards and AI video prompts. - Boundary markers: The workflow instructions do not define delimiters or specific 'ignore embedded instructions' warnings for the ingested novel content.
- Capability inventory: The skill has permissions to write to the file system (creating project directories, prompts, and text files) and execute media processing commands.
- Sanitization: There are no explicit steps for sanitizing or escaping the content extracted from user novels before it is passed to LLM-driven storyboard and prompt generation components.
Audit Metadata