edl-edit
Pass
Audited by Gen Agent Trust Hub on Aug 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides templates for executing the
video-agentCLI tool viauv runto perform rendering and transcription tasks. These commands are integral to the defined workflow. - [DYNAMIC_EXECUTION]: The EDL schema allows raw
ffmpegfilter strings in properties likeaudio_fixandgrade. These values are passed to the rendering engine at runtime, providing a mechanism for dynamic parameter injection into the processing pipeline. - [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to derive edit points from source media transcripts. As these transcripts represent external, untrusted data, they constitute an attack surface where adversarial audio content could attempt to influence agent behavior.
- Ingestion points: Word-level transcripts from
transcribe src.mp4are used by the agent to construct the EDL (SKILL.md). - Capability inventory: Execution of
video-agentshell commands (SKILL.md). - Boundary markers: The instructions do not define delimiters or protective prompts when processing transcript text.
- Sanitization: Transcript content is not sanitized before analysis.
Audit Metadata