ai-media-quality-review
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The bundled Python script executes
ffmpegandffprobe. The implementation uses list arguments forsubprocess.runinscripts/inspect_media.py, adhering to security best practices by avoiding shell interpretation and preventing command injection vulnerabilities.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided media files to generate quality reports.\n - Ingestion points: The
scripts/inspect_media.pyscript accepts a media file path as input.\n - Boundary markers: The agent is provided with specific, structured review instructions in
SKILL.mdto guide its evaluation.\n - Capability inventory: The script
scripts/inspect_media.pyusessubprocess.runto callffprobeandffmpegfor metadata extraction and artifact creation, and writes JSON and image files to the local disk.\n - Sanitization: External tool output is parsed using
json.loadsand specific regular expressions to ensure only expected data is processed.
Audit Metadata