deepgram-js-speech-to-text
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documents interfaces for processing external audio data which constitutes a potential injection surface.
- Ingestion points: The methods
transcribeUrlandtranscribeFileinSKILL.mdingest external audio URLs and local file streams. - Boundary markers: No specific boundary markers or "ignore instructions" prompts are included in the code examples, as they are standard SDK implementations.
- Capability inventory: The SDK performs network requests to the Deepgram API and can read local files via
fs.createReadStream. - Sanitization: The examples demonstrate raw output logging; downstream applications should sanitize transcription results before further processing.
- [EXTERNAL_DOWNLOADS]: The skill provides instructions to install related components using
npx skills add deepgram/skills. - Evidence: The command in
SKILL.mdreferences thedeepgram/skillsrepository. - Context: As this is the official repository of the skill's author (Deepgram), it is considered a legitimate vendor resource for cross-language product knowledge.
Audit Metadata