video-prompt-engineer
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data (descriptions, images) and reads local project files from
~/.gen-ai/projects/style/to generate text prompts. This creates a surface for indirect prompt injection where instructions hidden in the processed data could influence the agent's behavior. The risk is minimized as the skill is restricted to text output and explicitly forbidden from executing CLI commands. - Ingestion points: User-provided text and images; local style guide files (manifest.json, STYLE.md, etc.) in
~/.gen-ai/projects/style/<slug>/. - Boundary markers: Final prompts are delivered in fenced code blocks.
- Capability inventory: File read access to application-specific project paths; no file-write or shell execution capabilities.
- Sanitization: No explicit sanitization or filtering of external content is described.
- [EXTERNAL_DOWNLOADS]: The skill references official Picsart infrastructure, including API endpoints at
api.picsart.comand MCP servers atmcp.picsart.io. These references are part of the intended integration with the vendor's ecosystem. - [SAFE]: The skill instructions provide strong operational constraints, explicitly prohibiting the agent from executing CLI commands or generating video content, focusing purely on text-based prompt engineering.
Audit Metadata