Speech to Text Transcription
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using
ffmpegandwhisperfor media processing and transcription.\n- [EXTERNAL_DOWNLOADS]: The agent is directed to download media files from user-specified URLs for transcription, which involves making network requests to untrusted sources.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted audio and video content from local and remote sources, which creates a potential surface for indirect prompt injection attacks.\n - Ingestion points: Local media files and remote URLs processed as described in
SKILL.md.\n - Boundary markers: None identified; transcription results are processed and displayed directly.\n
- Capability inventory: Uses
ffmpeg(subprocess execution),whisper(subprocess execution), andcurl(network operations).\n - Sanitization: No specific sanitization or filtering of the transcribed text is mentioned before it is processed or summarized by the agent.
Audit Metadata