editor-camera-directing
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes project transcripts and metadata to identify visual beats for camera movements. Ingesting this external data creates an attack surface where malicious content in a transcript could attempt to influence the agent. 1. Ingestion points: Video transcripts and semanticBrief scenes found in SKILL.md. 2. Boundary markers: No delimiters or isolation instructions are present to separate transcript content from agent instructions. 3. Capability inventory: Commands are restricted to video editing tools including element_applyCameraMove, element_clearCameraMove, and frame inspection. 4. Sanitization: The skill does not define methods for sanitizing or escaping transcript text.
- [COMMAND_EXECUTION]: The skill utilizes official vendor CLI tools and MCP commands from @scrollmark/cli. These executions are scoped to the intended functionality of video editing and originate from the skill author scrollmark.
Audit Metadata