Good questions.
Plain answers.

Public video, Shorts, youtu.be, playlist, channel ID, handle, custom channel, and legacy user links are supported. Private, deleted, members-only, or otherwise unavailable videos cannot be fetched.

Yes. Paste a channel or playlist link and a Choose videos instead button appears beside Get Transcripts. It opens the full list with thumbnails, titles, and dates; tick the ones you want, filter a channel to only its videos or only its Shorts, and only the ticked ones are fetched. Pasted lists open the same page with everything ticked so you can untick. The browser extension offers the same choice on channel and playlist pages.

ScriptHaul uses caption tracks YouTube exposes for the video: creator-uploaded captions or automatic captions. It does not transcribe audio and never downloads the video or audio file.

Yes. Open the transcript page and use “Refresh this transcript”: the captions are fetched again from YouTube and the cached copy is replaced. A refresh counts as one new video against the daily allowance. Automatic captions are also re-checked once a month on their own, so a manual track a creator adds later is picked up without asking.

One job covers up to 500 videos, newest first — that keeps a single request from monopolizing the free service. When a channel or playlist has more, the finished job page shows exactly which slice it covered and offers a “Get Transcripts for the next 500” button that picks up where it stopped.

Yes. Every download is plain text — .txt, .srt, .vtt, .md, or .json — and a bulk ZIP contains only those text files plus two manifests. There is nothing executable, no installer, and no script; you can open any file in a plain text editor and read exactly what it contains. For verification, manifest.json and manifest.csv in every bulk ZIP list a SHA-256 checksum for each transcript file, so you can confirm nothing was altered after the archive was built.

For reading, effectively yes: anything already in the permanent cache serves without meaningful limits, and that share of traffic grows every day. Fetching captions nobody has requested before is paced so one person cannot monopolize the service: a job covers up to 500 videos (with a one-click “next 500” continuation), and each person can start 10 jobs and fetch 2,000 new videos per day, resetting at midnight UTC. There is no monthly quota. The only thing that ever pauses new fetching briefly is the shared fetch balance being refilled or a YouTube-side outage — cached transcripts keep working through both, and queued jobs resume on their own.

Every successful caption fetch is cached permanently. If the requested track is already cached, it can be served without a bot check, queue, or new request to YouTube.

New caption requests need pacing. A cold job processes videos steadily and reports each result live. You can bookmark the job page and return later; it does not restart when you refresh.

TXT comes three ways: Clean removes timestamps, Timestamped adds a time to each caption line, and Both-in-one puts clean text first, then a timestamped copy. SRT and VTT are standard subtitle files. Markdown is a bulleted, timestamped outline ready for Notion or Obsidian, and JSON carries each timed line as data for scripts and pipelines.

Manual captions in your language are preferred, then automatic captions and available translations. If fallback is enabled, the best available caption track can be used and its language/source is recorded in the manifest.

One transcript file per successful video, named by date and title, plus manifest.csv and manifest.json. The manifests record each video, its status, the caption track used, a SHA-256 checksum of its file, and any error class — so missing items never vanish silently.

Yes — the ScriptHaul extension on the Chrome Web Store grabs the transcript for the video you are watching — or the whole channel — with one click from the toolbar or right-click menu. Same free service, same limits, no account.

Yes. ScriptHaul runs a free MCP (Model Context Protocol) server at https://scripthaul.com/mcp — no API key, no account, nothing to configure. In Claude: Settings → Connectors → Add custom connector, paste the URL, authentication “None”. In Claude Code: claude mcp add --transport http scripthaul https://scripthaul.com/mcp. Most other MCP clients accept the URL directly, and it is listed on the official MCP registry and Smithery as ScriptHaul. Assistants get four tools: fetch a video’s transcript, list a channel’s or playlist’s videos, read a channel’s or playlist’s latest uploads, and produce a bulk ZIP download link. Cached transcripts are unlimited; uncached fetches share the usual daily allowance.

No captions means the video exists but exposes no caption track. Unavailable means YouTube would not make the video playable—often because it is private, deleted, members-only, or region restricted.

Cached transcripts keep working. Queued jobs hold their place and resume on their own once the fetch route recovers or the shared fetch balance is topped up. The status page shows which situation applies.

ScriptHaul does not grant rights to anyone's words. Public availability is not a copyright license. You are responsible for permission, attribution, fair-use analysis, and any other obligations that apply to your use.

No. ScriptHaul is an independent utility and is not affiliated with, endorsed by, or sponsored by YouTube or Google.