azure-ai-transcription-py
Warn
Audited by Gen Agent Trust Hub on Feb 14, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- Unverifiable Dependencies (MEDIUM): The skill recommends installing 'azure-ai-transcription', which is not a standard official Azure SDK package (the official package is 'azure-cognitiveservices-speech'). This poses a risk of dependency confusion where a malicious actor could register the name to execute code on the user's system.
- Indirect Prompt Injection (LOW): 1. Ingestion points: The skill fetches and processes untrusted audio data from URLs provided in 'content_urls' (SKILL.md). 2. Boundary markers: No boundary markers or 'ignore' instructions are present to delimit untrusted transcription output. 3. Capability inventory: The skill transcribes audio to text; while no direct shell execution or file-write capabilities are shown, the resulting text could influence the agent's logic. 4. Sanitization: No evidence of sanitization or validation of the transcribed content is provided.
Audit Metadata