菜单
Transcribe audio, video, YouTube, Vimeo, and generic media URLs with iapp-transcriber or the hosted Skills runtime. Supports OpenAI GPT-4o transcription, OpenAI diarization, ElevenLabs Scribe v2, DeepGram, chunking, source metadata, subtitles, and JSON outputs.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Merge a GitHub pull request, merge when green, use a merge queue, or decide whether a pull request is mergeable. Use only for explicit merge intent, not ordinary review.
Run hosted browser automation and page extraction through the Skills runtime.
Run hosted deep research with parallel search, synthesis, citations, source notes, and downloadable report artifacts.
Generate images using OpenAI, Minimax, or Gemini through the hosted Skills runtime with provider-cost pricing.
Create and manage grouped tmux sessions where each session focuses on its own window and its own folder. Knows workspace layout, naming conventions, and both multi-project and multi-agent patterns.
Take a high-level goal and drive it to done — break it into tasks via the todos CLI, open a goal thread in the project's conversations channel, create a plan in todos, generate the tasks under that plan, then execute them one by one until complete. Use for 'execute this goal', 'break down and do X', 'set a goal and run it'.
| name | transcript |
| description | Transcribe audio, video, YouTube, Vimeo, and generic media URLs with iapp-transcriber or the hosted Skills runtime. Supports OpenAI GPT-4o transcription, OpenAI diarization, ElevenLabs Scribe v2, DeepGram, chunking, source metadata, subtitles, and JSON outputs. |
Create transcripts from local audio/video files or media URLs. Use this skill when the user asks to transcribe, caption, diarize, summarize, or package spoken audio/video content.
skills run transcript, needs remote execution, or has only SKILLS_API_KEY configured.iapp-transcriber when you are on this machine and need direct access to local files, YouTube/Vimeo/generic yt-dlp sources, transcript DB records, MCP tools, comments, exports, or OpenLoops follow-up workflows.transcriber when installed, or bun run src/cli/index.ts from your ~/Workspace/hasnaxyz/internalapp/iapp-transcriber checkout.skills run transcript --source ./meeting.mp3 --title "Design review" --provider openai
skills run transcribe --source https://www.youtube.com/watch?v=... --provider openai --diarize
Poll hosted runs with skills runs status <run-id> and download outputs with skills exports download <run-id>.
transcriber transcribe ./meeting.mp3 --provider openai --json
transcriber transcribe https://www.youtube.com/watch?v=... --provider openai --model gpt-4o-transcribe --json
transcriber transcribe ./meeting.mp3 --provider openai --diarize --json
transcriber export <transcript-id> --format srt --output captions.srt
Local provider defaults:
openai: default, uses gpt-4o-transcribe; --diarize uses gpt-4o-transcribe-diarize.elevenlabs: uses scribe_v2, supports diarization and keyterms.deepgram: uses Nova-3, supports diarization.Local requirements:
yt-dlp for remote media URLs. Set YTDLP_PATH if needed.ffmpeg/ffprobe; the local app bundles npm ffmpeg/ffprobe and also respects FFMPEG_PATH and FFPROBE_PATH.Inspect source metadata first for URLs:
transcriber info <url> --json
Download audio when the user asks to keep media:
transcriber download <url> --format mp3 --json
Transcribe with JSON for automation:
transcriber transcribe <path-or-url> --provider openai --json
Export or post-process:
transcriber get <id> --json
transcriber export <id> --format txt --output transcript.txt
transcriber summarize <id>
For repeat work, create OpenLoops command loops around JSON-producing commands, for example transcriber feed check --json --dry-run.
TRANSCRIBER_ALLOW_PRIVATE_URLS=1 only for trusted internal sources.--json for scripts and OpenLoops so failures include a structured transcript record and nonzero exit code.