photoreal-video-production
Pass
Audited by Gen Agent Trust Hub on Nov 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill references local shell commands and Python scripts (e.g.,
init-project.sh,make.py) which are part of a shared 'video-production' pipeline. These are standard workflow components for project initialization and asset management. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data including client photos, voiceover lines, and product details to generate video prompts.
- Ingestion points: Processes supplied photos for character identities and voiceover text for emotional context in
references/characters.mdandreferences/clip-prompt.md. - Boundary markers: Uses a structured JSON blueprint (
references/still-blueprint.md) and explicit asset referencing (e.g.,@image1) to isolate ingredients. - Capability inventory: The skill writes prompt files to the local file system and executes local production scripts.
- Sanitization: Includes explicit instructions to adhere to platform safety filters and avoid generating prohibited content (e.g., intimacy filters mentioned in
references/clip-prompt.md). - [SAFE]: No remote code downloads, obfuscated content, or unauthorized exfiltration patterns were identified. The technical instructions align with the stated purpose of professional video production workflows.
Audit Metadata