메뉴
Summarize or extract text/transcripts from URLs, podcasts, and local files (great fallback for “transcribe this YouTube/video”).
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when the user asks to use a consumer integration that is not set up yet: WhatsApp as Me, Email, Google Workspace, Apple Notes, Apple Reminders, Telegram as Me, Google Maps Search, or creative audio. Route here only when the right integration exists but is blocked by missing login, OAuth, QR pairing, permissions, local dependency setup, configuration, or API credentials, and the response should guide setup in product language instead of dumping CLI commands.
Use for Google Workspace requests tied to Gmail, Google Calendar, Drive, Docs, Sheets, or Contacts, especially when the user explicitly mentions Google or needs cross-surface Google account access. Prefer this over generic email skills when the task is clearly in the Google ecosystem.
Use when the user asks to triage messages, emails, chats, inboxes, unread messages, recent messages, open replies, who they owe replies to, outreach, or reply drafting across one or more channels such as WhatsApp, Telegram, Gmail/email, Slack, Signal, iMessage, browser portals, or files. For broad sweeps, first produce a scoped prioritized triage list; for one known target, use single-thread draft mode. Always require user approval before any external reply/send.
Use for generic email tasks on connected mail accounts: search, read, summarize, draft, reply, forward, send, and organize email over IMAP/SMTP. This is the better match for normal mailbox work that is not specifically a Gmail or Google Workspace request.
Use for Telegram-as-me requests on this Mac: reading, sending, replying, or waiting as the user's real Telegram account. Do not use it for the normal Telegram bot channel, BotFather setup, or generic bot onboarding.
Send WhatsApp messages to other people or search/sync WhatsApp history via the wacli CLI (not for normal user chats).
| name | summarize |
| description | Summarize or extract text/transcripts from URLs, podcasts, and local files (great fallback for “transcribe this YouTube/video”). |
| homepage | https://summarize.sh |
| metadata | {"openclaw":{"emoji":"🧾","requires":{"bins":["summarize"]},"install":[{"id":"brew","kind":"brew","formula":"steipete/tap/summarize","bins":["summarize"],"label":"Install summarize (brew)"}]}} |
Fast CLI to summarize URLs, local files, YouTube links, podcasts, and audio/video files.
Use this skill immediately when the user asks any of:
yt-dlp needed)summarize "https://example.com"
summarize "/path/to/file.pdf"
summarize "https://youtu.be/dQw4w9WgXcQ" --youtube auto
summarize "/path/to/podcast.mp3" --transcriber auto
Best-effort transcript/extraction:
summarize "https://youtu.be/dQw4w9WgXcQ" --youtube auto --extract
summarize "/path/to/podcast.mp3" --transcriber auto --extract
If the user asked for a transcript but it’s huge, return a tight summary first, then ask which section/time range to expand.
Set the API key for your chosen provider:
OPENAI_API_KEYANTHROPIC_API_KEYXAI_API_KEYGEMINI_API_KEY (aliases: GOOGLE_GENERATIVE_AI_API_KEY, GOOGLE_API_KEY)Default model is auto if none is set; pass --model <provider/model> only when a specific provider is needed.
--length short|medium|long|xl|xxl|<chars>--max-output-tokens <count>--extract (print extracted content/transcript and exit)--json (machine readable)--firecrawl auto|off|always (fallback extraction)--youtube auto|web|no-auto|yt-dlp|apify--transcriber auto|whisper|parakeet|canary (audio transcription)--video-mode auto|transcript|understand--slides, --slides-ocr, --slides-max <count> (slide-heavy videos)Optional config file: ~/.summarize/config.json
{ "model": "auto" }
Optional services:
FIRECRAWL_API_KEY for blocked sitesAPIFY_API_TOKEN for YouTube fallback