ai-video-generator
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes
subprocess.runto execute FFmpeg for video processing. While it follows best practices by using argument lists rather than shell strings, it dynamically constructs an FFmpeg filter string (subtitle_filter) using f-string interpolation of a filename. If filenames were to contain certain special characters, it could potentially cause the filter to fail or behave unexpectedly. - [PRIVILEGE_ESCALATION]: Setup instructions include a
sudo apt install ffmpegcommand. While FFmpeg is a standard dependency for video manipulation, the requirement for root access during environment setup is a privilege escalation finding that should be noted. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The
topicparameter in thegenerate_scriptandgenerate_videofunctions (found inSKILL.md) accepts arbitrary user input. - Boundary markers: None. The user-provided
topicis directly interpolated into a f-string prompt passed to the Anthropic API. - Capability inventory: The skill possesses the capability to perform network operations (
requests), file system writes (open), and system command execution (subprocess.run). - Sanitization: There is no evidence of input validation, sanitization, or escaping of the
topicvariable before it is used to generate the video script, making it susceptible to prompt injection attacks where a user could override the intended script format or content.
Audit Metadata