deepgram-transcribe
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides a bash script that utilizes curl for API requests and an inline Python script for processing JSON results.
- [DATA_EXFILTRATION]: Audio data is transmitted to api.deepgram.com, which is a well-known technology service. This operation is essential for the skill's purpose and uses environment variables for secure authentication.
- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it transcribes audio data that is then returned to the agent context. 1. Ingestion point: local audio files processed by the curl command. 2. Boundary markers: none. 3. Capability inventory: usage of curl and python3 to extract and display transcripts. 4. Sanitization: the script does not sanitize the transcribed text before output.
Audit Metadata