claude-video-audio
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external media files that could contain malicious payloads in metadata or subtitle streams.\n
- Ingestion points: Processes user-provided media files via the
$INPUTvariable across multiple FFmpeg and Bash operations defined inSKILL.md.\n - Boundary markers: Absent. The instructions do not include specific delimiters or warnings to ignore instructions embedded within the processed data.\n
- Capability inventory: The skill has access to the
Bashtool, allowing for extensive file system and potentially network operations throughffmpegand shell scripts.\n - Sanitization: Absent. There is no explicit instruction to sanitize or validate the content or filenames of the ingested data.\n- [COMMAND_EXECUTION]: The skill utilizes shell commands that incorporate user-controlled input (
$INPUT,$OUTPUT). If the execution environment does not adequately sanitize these variables, it could lead to command injection attacks.\n- [EXTERNAL_DOWNLOADS]: The skill relies on the external utilityauto-editor, which is not a standard system tool and introduces a dependency on the host environment's pre-installed software.
Audit Metadata