Any Social Video → Ready-to-Use Text, in Your Language

Paste a link from YouTube, Instagram, X / Twitter, TikTok, Facebook, Vimeo, DailyMotion, or Loom — get a clean verbatim transcript with an AI summary and key points, often under two minutes. The transcript stays in the spoken language so hooks and quotes hold up under scrutiny; the summary auto-translates to your chat language so you can scan, share, or paste it into your workflow on the spot.

What it does

Turns any public social-media video link into a downloadable transcript document with three sections: a Summary (2–4 sentences), Key Points (5–7 bullets pulling out the substantive claims, named entities, and conclusions), and the full Transcript in the spoken language, paragraphed for readability.

Section headers and the summary localize automatically to whatever language you're writing in — English, Chinese (Simplified and Traditional), Japanese, Korean, Spanish, French, German, Italian, Portuguese, Dutch, Arabic, and more. The transcript itself is never translated and never rewritten — viral hooks, on-record quotes, and platform-specific phrasing keep their exact wording.

For short clips (≤ 2,000 characters of transcript — typical of TikToks, Reels, and Shorts), the full text is also inlined in the chat reply. Longer videos return Summary + Key Points with a one-line file pointer, so a 30-minute podcast doesn't dump a wall of text into your conversation.

What's different about this skill

Most transcription tools solve one problem: speech to text. This skill solves the whole workflow — from raw link to a document you can actually use — without requiring you to pick the right platform, set the language, or clean up the output yourself.

Verbatim transcript, translated summary — never mixed up. The transcript is never rewritten or translated. That means a viral hook stays quotable, a CEO's exact words stay on-record, and a subtitled clip doesn't get paraphrased into something that sounds almost right. The summary and key points, on the other hand, come back in whatever language you're writing in — no extra prompt needed. Most tools make you choose: get the original, or get a translation. This one keeps both, cleanly separated.

Eight platforms, one prompt. YouTube-only tools are common. Tools that also handle Instagram, TikTok, X, Facebook, Vimeo, DailyMotion, and Loom — and auto-detect which one you've pasted — are not. No "which platform?" follow-up, no plugin-switching, no copy-pasting the URL into a different tool for Reels vs. Shorts.

Output that fits the video length. A 45-second TikTok and a 40-minute podcast are not the same problem. Short clips inline the full transcript in chat so you can read and copy without opening a file. Long videos return Summary + Key Points in chat and save the full text to a downloadable file — so a dense lecture doesn't dump thousands of words into your conversation. The threshold is automatic, not a setting you have to configure.

Three output formats, one ask. Ask for timestamps and you get paragraph-level [mm:ss] markers in Markdown. Ask for subtitles or SRT and you get caption blocks with distributed timing, ready for a video editor. Say nothing and you get clean plain text. No mode-switching, no separate export step.

AI language detection, built in. You don’t need to choose a language before transcription. The skill uses AI to identify the spoken language in the video and export an accurate transcript in the original language. English, Chinese, Japanese, Korean, Spanish, multilingual clips — the workflow stays the same: paste the link, get clean video transcription, summary, key points, and a ready-to-use transcript without manual setup.

Who it's for

• Creators and content teams clipping content for re-upload, building captions, or localizing campaigns across platforms
• Journalists and researchers quoting from clips, fact-checking viral moments, and pulling exact words without rewatching
• Students and podcast listeners turning lectures, talks, interviews, and long-form videos into searchable notes
• Brand, growth, and CRO teams extracting hooks and insights from audio-heavy posts and competitor content
• Anyone who wants the words but not the rewatch — TikToks playing under loud music, long YouTube tutorials, embedded X video clips, busy Reels, dense lectures

How it works

1. Paste any public link. Reels, IGTV, feed posts, Shorts, TikTok videos (including regional variants), Facebook Watch / Live replays / page posts, viral X clips, news segments, embedded videos, Vimeo clips, DailyMotion videos, Loom recordings — the skill auto-detects the platform and source language.
2. Tell it what you need if the default isn't right. "with timestamps" returns timestamped Markdown with [mm:ss] paragraph markers; "subtitles" or "SRT" returns a captions file ready for video editors; anything else returns clean plain text.
3. Get back a transcript document. Summary and Key Points appear inline in chat in your language; the full Transcript is saved to a downloadable file in the spoken language. Short videos also inline the full transcript for quick scanning.

Output formats

• Plain text (.txt) — clean prose without timing noise. Default for reading, summarizing, or feeding into another tool.
• Timestamped Markdown (.md) — paragraph-level [mm:ss] markers. Best for hunting a specific quote or anchoring back to the video.
• SRT subtitles (.srt) — sentence-level timing blocks ready for re-upload, with distributed timing so consecutive lines never overlap or zero out.

Language handling

Transcript stays verbatim in the spoken language so viral hooks, on-record quotes, and platform-specific phrasing remain exact for re-quoting, fact-checking, or re-upload. Summary, Key Points, and section headers render in your chat language — write to the skill in Chinese and get Chinese summaries of an English video; write in Japanese and get Japanese summaries of a Korean clip.

Source language is auto-detected from URL signals where possible: Bilibili, Douyin, Niconico, Naver, AfreecaTV, and similar mono-lingual domains lock to their native language; non-Latin handles on multi-lingual platforms route to the matching language. When signals are ambiguous, the skill defaults to the platform's dominant language and silently retries once in your language if the first attempt comes back empty.

Example

Input:

"Transcribe this and give me the key takeaways: https://www.youtube.com/watch?v=..."

Output (chat reply):

• Summary — 2–4 sentences in the user's chat language describing what the video covers.
• Key Points — 5–7 bullets pulling out the substantive claims, named entities, and conclusions.
• File pointer — Full transcript saved to transcript_<slug>.txt (localized to the user's language).

Output (downloadable file): metadata header + the same Summary and Key Points + the full verbatim Transcript in the spoken language, paragraphed for readability.

If the user adds a specific question — "what did they say about pricing?" — the answer appears as a short verbatim quote at the top of the reply, before the Summary section.

Input formats

• Supported platforms: YouTube, Instagram (Reels, feed posts), X / Twitter, TikTok (including regional variants), Facebook (Watch, Live replays, Reels, page posts), Vimeo, DailyMotion, Loom
• Public links only — private posts, region-locked content, login-required pages, and live streams can't be reached
• Any spoken language — auto-detected, with a single silent retry on empty result before the skill asks for guidance
• Output language — follows your chat language; transcript stays in source spoken language and is never translated
• Out of scope: local files on your computer (use a separate audio-to-text skill), articles or silent web pages, live streams, translation of the transcript itself