The Agent Skills Directory

[SAFE]: The skill documents interactions with api.venice.ai, which is the official API domain of the vendor (veniceai).
[DATA_EXPOSURE]: The skill uses environment variables like $VENICE_API_KEY for authentication examples, which is a standard best practice for managing secrets and prevents hardcoding credentials.
[EXTERNAL_DOWNLOADS]: The documentation mentions standard, well-known tools and libraries such as the openai Node.js SDK, ffmpeg, and pydub. These are appropriate for audio processing and API integration.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates processing untrusted audio data into text. While this text is returned to the agent, the skill itself does not define any downstream automated execution or unsafe processing of that data.
Ingestion points: The file parameter in the POST /api/v1/audio/transcriptions endpoint (SKILL.md).
Boundary markers: None specified for the audio content.
Capability inventory: Network requests (POST) and file reads (fs.createReadStream) (SKILL.md).
Sanitization: Not applicable as the skill performs transcription and returns raw output.

venice-audio-transcription