azure-speech-to-text
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill converts audio and video files into text transcripts that are subsequently processed by the agent. This creates a surface for indirect prompt injection, where malicious instructions spoken in the source media could be interpreted as valid commands by the agent when it reads the resulting transcript.
- Ingestion points: The
azure_stttool accepts external audio files via theinput_pathparameter. - Boundary markers: There are no instructions or delimiters defined to separate the transcribed content from system instructions or to warn the agent against following instructions embedded in the audio.
- Capability inventory: The skill performs file system reads and network operations (transmitting audio data to Azure APIs).
- Sanitization: No mechanisms for filtering, sanitizing, or validating the semantic content of the transcribed text are implemented.
Audit Metadata