Loom Archiver

Archive the Loom videos you created — video, transcript, and a resumable ledger — to a folder you control.

Loom has no bulk export. Downloads are one video at a time, gated to paid plans, and transcripts cannot be exported at all. This tool walks your library over Loom's own API and saves everything locally.

Your credentials go only to Loom. You sign in through a real Chrome window (Google SSO included); the tool never sees your password, only the resulting session cookie, stored 0600 in your platform config directory. That cookie is sent to Loom itself — exactly as your browser does — and nowhere else. There is no backend, no telemetry, and no third party in the path.

Proven on a 1,658-video / 314 GB library.

New here? docs/WALKTHROUGH.md is a step-by-step guide from installation to a verified archive, with the output you should expect at each stage and what to do when something fails. The rest of this README is the reference.

Install

pipx install 'loom-archiver[auth]'
pipx run playwright install chrome

(pipx only puts the installed package's own command on your PATH, not playwright's — pipx run fetches Playwright into a throwaway environment just to run its installer. The Chrome build it downloads lands in a shared user-level cache, so the loom-archiver command finds it regardless of which environment triggered the download.)

ffmpeg is required and is not installed automatically:

Platform	Command
macOS	brew install ffmpeg
Debian/Ubuntu	sudo apt install ffmpeg
Windows	https://ffmpeg.org/download.html

Usage

loom-archiver auth      --dest ~/loom-archive   # sign in once, in a real browser
loom-archiver inventory --dest ~/loom-archive   # list what exists; downloads nothing
loom-archiver run       --dest ~/loom-archive   # download everything

run is resumable — Ctrl-C and rerun it. Progress lives in <dest>/loom_video_list_progress.csv.

Options: --floor <GB> (stop when free space drops below this, default 50), --delay <seconds> (pause between requests, default 0.5), --max-depth <n> (maximum folder nesting depth to walk, default 10), --folders-only (skip the main-library pass and discover only videos inside folders). --folders-only scopes discovery, not processing: against a destination that already has a ledger, videos left pending from an earlier run are still processed.

What you get

<dest>/
  main-library/         videos not in any folder
    <id>__<title>.mp4
    <id>__<title>.vtt   transcript, original cue format
    <id>__<title>.txt   transcript, plain text
  <Folder>/             one directory per Loom folder, nested as in Loom
  loom_video_list_progress.csv

Every run prints what it enumerated — loose count, folder count, per-folder counts — so you can see the shape of your archive rather than infer it.

Limitations

• Videos you created only. Videos shared with you are not archived.
• No workspace/admin archive. There is no "download everything in the org" mode.
• Downloaded MP4s do not include Loom features that only exist in their player (CTAs, chapters, filler-word removal).
• A small number of videos are served as DASH; those fall back to yt-dlp.
• Deeply nested folders are walked to depth 10 by default and will raise rather than silently truncate beyond that.

Credentials

Session state lives in your platform config directory:

• macOS: ~/Library/Application Support/loom-archiver/
• Linux: ~/.config/loom-archiver/ (or $XDG_CONFIG_HOME/loom-archiver/)
• Windows: untested. There's no Windows-specific path, so it falls back to the Linux logic above (typically ~\.config\loom-archiver\, which is not an idiomatic Windows location). Set LOOM_ARCHIVER_CONFIG_DIR to place it explicitly, e.g. %LOCALAPPDATA%\loom-archiver.

Override with LOOM_ARCHIVER_CONFIG_DIR. Treat auth_state.json like a password. Sign out of the Loom session when your archive is complete.

How it works

Loom's web app talks to a private GraphQL API; this tool uses the same operations your browser does — GetLoomsForLibrary to page through videos, GetPublishedFolders to walk folders, FetchVideoTranscript for captions, and GetVideoSource / GetVideoTranscodedUrl for media. Videos served only as CloudFront-signed HLS are fetched segment by segment and muxed with ffmpeg.

Legal

This tool accesses your own content, in your own account, with your own credentials. It does not redistribute anything, does not touch other people's data, and rate-limits itself by default.

Your use of Loom remains subject to your agreement with Loom/Atlassian — compliance with those terms is between you and them. Not affiliated with Loom or Atlassian.

License

MIT © 2026 Will Dent