GIF
Pass
Audited by Gen Agent Trust Hub on Apr 20, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it processes untrusted user data—specifically video files and search strings—which are then passed to shell commands and external APIs. * Ingestion points: Input video files (input.mp4) and search parameters (q=) in SKILL.md. * Boundary markers: Absent; commands do not include delimiters or 'ignore' instructions for processed content. * Capability inventory: Execution of ffmpeg and gifsicle via subprocess, and network requests via curl in SKILL.md. * Sanitization: Absent; the provided examples do not demonstrate validation or escaping of user-provided paths or queries.
- [COMMAND_EXECUTION]: Instructs the agent to execute local binaries for media conversion and optimization. * Evidence: Shell commands for ffmpeg and gifsicle in SKILL.md.
- [EXTERNAL_DOWNLOADS]: Fetches GIF search results and data from well-known services. * Evidence: Requests to api.giphy.com and tenor.googleapis.com in SKILL.md.
Audit Metadata