AgentSkillsCN

openai-transcribe

通过aivault能力JSON，借助OpenAI的语音转文本功能，对本地音频文件进行转录（技能运行时无需提供提供商API密钥）。

SKILL.md
--- frontmatter
name: openai-transcribe
description: Transcribe local audio files with OpenAI speech-to-text through aivault capability json (no provider API key in skill runtime).
compatibility: Requires `aivault` CLI and Node.js (runs via `npx tsx`).
credentials:
  - name: OPENAI_API_KEY
    description: OpenAI API key for speech-to-text transcription
    required: true

OpenAI Transcribe

This skill uses aivault to always keep your API keys safe.

Setup

Install aivault if it is not already available:

bash
curl -fsSL https://aivault.moldable.sh/install.sh | sh

Store your OpenAI API key in aivault. The credential and all openai/* capabilities are auto-provisioned from the built-in registry.

bash
aivault secrets create --name OPENAI_API_KEY --value "sk-..." --scope global

This skill uses openai/transcription.

Quick start

bash
npx -y tsx {baseDir}/scripts/transcribe.ts /path/to/audio.m4a

Common examples

bash
npx -y tsx {baseDir}/scripts/transcribe.ts /path/to/audio.wav
npx -y tsx {baseDir}/scripts/transcribe.ts /path/to/audio.wav --json

Notes

  • •Capability id is hard-coded to openai/transcription.
  • •Default model sent to upstream is whisper-1.
  • •The script uses aivault json so upstream JSON is parsed/validated by aivault; without --json it prints just the extracted transcript text.