Video Project Setup
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input and writes it to the local workspace context, creating an attack surface for indirect prompt injection.
- Ingestion points: Processes user-supplied creative objectives, scripts, and brand facts.
- Boundary markers: Includes specific instructions to maintain distinctions between
unknown,user-supplied,provider-observed, andapprovedstates. - Capability inventory: Uses
workspace-writepermissions to create or update the.video/project.mdfile. - Sanitization: Instructs the agent not to convert unverified statements into facts and to summarize differences before updating existing contexts.
Audit Metadata