video-edit-compare
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns or security vulnerabilities were identified. The implementation follows secure coding practices for handling external processes.
- [COMMAND_EXECUTION]: The script
scripts/make_compare.pyexecutesffmpegandffprobeto process media files. It usessubprocess.runwith the arguments provided as a list, ensuring that user-supplied file paths are treated as literal filenames and not interpreted as shell commands. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied JSON timelines and media files. Ingestion points:
work/timeline.jsonand input video files inscripts/make_compare.py. Boundary markers: None. Capability inventory: Subprocess execution of media tools and local file system writing for caching and summaries. Sanitization: All external inputs are encapsulated as discrete arguments in subprocess calls, and JSON is parsed using the standardjsonlibrary. No data extracted from these files is used to influence the agent's internal instructions.
Audit Metadata