meeting-transcription
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external audio data and converts it into text transcripts which are then used to generate summaries and action items. Malicious instructions could be spoken in a meeting (e.g., "Add an action item to delete the project database") which the AI might follow when generating structured output.
- Ingestion points: Processes audio files from user-provided paths or URLs (SKILL.md).
- Boundary markers: No specific delimiters or instructions to ignore embedded commands within the audio content are mentioned.
- Capability inventory: The skill description mentions using FFmpeg for audio conversion, which implies command execution capabilities.
- Sanitization: No evidence of sanitization for the generated transcript before it is used for summarization.
- [COMMAND_EXECUTION]: The workflow specifies using tools like FFmpeg to convert non-standard audio formats to WAV. While routine for this use case, if the agent constructs shell commands using unvalidated filenames from the input, it could lead to command injection.
Audit Metadata