🌐 Landing page · 📦 Download latest · 🐛 Report issue
A privacy-respecting Chrome extension that lets you download videos, images, and audio from YouTube, Twitter/X, Instagram, TikTok, Reddit, Facebook, Twitch, and Bilibili — with a polished popup UI, a floating hover overlay, per-site toggles, dark mode, settings sync, and a 3-tier download fallback that handles restrictive CDNs.
No upfront host permissions — GrabIt asks for access only when you first use it on a new site. Revoke anytime in Settings → Sites.
All extraction and downloading happen locally in your browser. No third-party servers, no telemetry, no account required (beyond being signed into each site as needed).
- 8 site extractors with a unified registry pattern — easy to add more
- 3-tier download fallback that handles referer-gated CDNs:
chrome.downloads.download()(direct, fastest)- Service-worker
fetch()+ Blob URL download (bypasses referer via DNR rules) - Open in new tab (last resort)
- DASH + HLS manifest parsing for adaptive streams (Reddit, Twitch VODs)
- DeclarativeNetRequest rules that inject
Referer/Originheaders for CDNs that require them (video.twimg.com, v.redd.it, tiktokcdn, etc.)
- Modern popup with brand header, tab-info card, loading skeleton, media cards with thumbnails, variant grid, recent downloads list, and toast notifications
- Hover overlay — a floating "GrabIt" button that appears on supported sites, with a popover showing all available quality variants; uses Shadow DOM so site styles never leak in
- Full options page with sidebar nav (General / Sites / Downloads / Overlay / Advanced / About), segmented controls, color picker, range sliders, toggle switches
- Dark mode — Auto / Light / Dark with system-preference detection
- 8 accent colors to choose from
- Settings sync via
chrome.storage.sync— your preferences follow you across signed-in Chrome instances - Per-site enable/disable — toggle each extractor independently
- Filename templates with variables
{site} {title} {quality} {ext} {author} {date} - Desktop notifications on completion and errors
- Recent downloads list (last 50) with status indicators
- Context menu integration ("GrabIt: detect downloadable media on this page")
- SPA navigation detection — overlay re-mounts on YouTube/Twitter route changes
- Reduced-motion support for accessibility
| Site | URL pattern | Notes |
|---|---|---|
| YouTube | youtube.com/watch, /shorts, /embed; youtu.be/* |
Ciphered streams fall back to the yt-bridge MAIN-world script |
| Twitter / X | twitter.com/*/status/*, x.com/*/status/* |
Uses bot UA + cookie; referer-bypassed via DNR |
instagram.com/reel/*, /p/*, /tv/* |
Requires login (cookies); private/protected posts unsupported | |
| TikTok | tiktok.com/@*/video/* |
Scrapes universal data blob; signed URLs expire ~1h |
reddit.com/r/*/comments/* |
Uses DASH or HLS master playlist; falls back to direct MP4 | |
facebook.com/watch, /permalink, /reel, /*/videos |
Requires login; scrapes playable_url from embedded JSON |
|
| Twitch | twitch.tv/*/clip/*, twitch.tv/videos/* |
Clips exposed as direct MP4s; VODs via m3u8 |
| Bilibili | bilibili.com/video/BV* |
Uses public API; DASH video+audio are separate variants |
grabit/
├── manifest.json MV3 manifest, permissions, DNR ruleset
├── background/
│ ├── background.js Service worker: IPC, download orchestration, 3-tier fallback
│ └── dnr_rules.json Static DNR rules (referer/origin spoofing per CDN)
├── content/
│ ├── content.js Content script: SPA detection, overlay mount, IPC bridge
│ ├── hover-overlay.js Floating download button widget (Shadow DOM)
│ └── hover-overlay.css Overlay host styles
├── extractors/
│ ├── registry.js Extractor registration + dispatch
│ ├── youtube.js YouTube (ytInitialPlayerResponse scrape)
│ ├── twitter.js Twitter (bot UA HTML scrape)
│ ├── instagram.js Instagram (video_versions JSON scrape)
│ ├── tiktok.js TikTok (universal data blob scrape)
│ ├── reddit.js Reddit (.json API + DASH/HLS parse)
│ ├── facebook.js Facebook (playable_url JSON scrape)
│ ├── twitch.js Twitch (clip MP4 + VOD m3u8)
│ └── bilibili.js Bilibili (public API + DASH)
├── lib/
│ ├── utils.js Logging, URL deobfuscation, settings, storage, helpers
│ ├── dash-hls.js HLS master + DASH MPD parser
│ ├── restriction-bypass.js CDN→origin map, DNR rule builder, fetch fallback
│ └── yt-bridge.js YouTube signature bridge (MAIN-world script)
├── popup/
│ ├── popup.html 380px-wide popup markup
│ ├── popup.css Modern card UI with dark mode
│ └── popup.js Popup controller: scan, render, download
├── options/
│ ├── options.html Full-page options UI (sidebar nav)
│ ├── options.css Card system, toggles, segmented controls
│ └── options.js Options controller with debounced save
├── icons/
│ ├── icon16.png Generated by scripts/gen_icons.py
│ ├── icon48.png
│ └── icon128.png
├── AGENT_MEMORY.md Architecture notes for AI agents resuming work
├── README.md This file
├── LICENSE MIT
└── .gitignore
User clicks toolbar icon
│
▼
popup.js boots
│
│ chrome.runtime.sendMessage({type:'scanActiveTab'})
▼
background.js
│
│ runExtractorForTab(tab, settings)
▼
registry.js → site-specific extractor
│
│ fetch(url, {credentials:'include'}) → parse HTML/JSON
▼
returns MediaResult[] { id, title, variants[] }
│
▼
popup.js renders cards
│
│ user clicks a variant
▼
background.js.startDownload(payload)
│
▼
Tier 1: chrome.downloads.download()
│ └─ if needsReferer or fails →
▼
Tier 2: fetchWithCredentials() → Blob → URL.createObjectURL → chrome.downloads.download()
│ └─ if fails →
▼
Tier 3: chrome.tabs.create({url})
content.js (document_idle)
│
│ loadSettings() → scheduleOverlayMount()
▼
fetch(hover-overlay.js) → inject as <script>
│
▼
window.GrabItOverlay.mount({settings, onDownload, getMedia})
│
│ patches history.pushState/replaceState for SPA detection
▼
user clicks floating button → popover opens
│
│ user picks a variant
▼
chrome.runtime.sendMessage({type:'download', payload})
- Download the latest
grabit.zipfrom releases (or clone this repo) - Unzip it to a permanent folder (Chrome needs it to stay in place)
- Open Chrome and navigate to
chrome://extensions/ - Enable Developer mode (top-right toggle)
- Click Load unpacked and select the unzipped
grabit/folder - The GrabIt icon will appear in your toolbar — pin it for easy access
- Visit a supported site (e.g. a YouTube video) and click the GrabIt icon
- First time only: the popup will ask you to grant host permission for that site (no upfront permissions are requested at install time). Click "Enable for this site" — Chrome shows its standard permission prompt. Grant it and the popup will re-scan automatically.
The PNG icons are generated from scripts/gen_icons.py:
pip install pillow
python scripts/gen_icons.pyA GitHub Actions workflow (.github/workflows/build.yml) runs on every push to main:
- Builds a fresh zip of the extension
- Replaces the
latestrelease asset (grabit.zip) — the previous zip is clobbered (effectively deleted) - On major version bumps (e.g. v1.x → v2.x), snapshots the previous
latestzip into a permanentv{N}.x-archiverelease before replacing it
So users always have:
- Latest — the most recent build, replaced on every commit
v{N}.x-archivereleases — one per major version, never modified (e.g.v1.x-archivecontains the last v1.x build, kept forever)
- Navigate to a supported site (e.g. a YouTube video)
- Click the GrabIt toolbar icon
- The popup shows the detected media with all available quality variants
- Click any variant to start the download — a toast confirms the tier used
- Recent downloads appear at the bottom of the popup
- On any supported site, a floating "GrabIt" button appears in the corner
- Click it to see all available media and quality variants
- Pick a variant to download immediately
Right-click anywhere on a supported page → GrabIt: detect downloadable media on this page → the first detected media's best variant will download automatically.
Open the options page (right-click the toolbar icon → Options, or click Settings in the popup footer) to configure:
- Theme (Auto / Light / Dark) and accent color
- Per-site enable/disable
- Default download quality
- Download subfolder and filename template
- Notification preferences
- Hover overlay position, opacity, and visibility
- Restriction bypass (DNR rules)
- Max parallel downloads
- Debug logging
Settings sync across your signed-in Chrome instances via chrome.storage.sync.
GrabIt uses optional host permissions — no site access is granted at install time. When you first click the GrabIt icon on a supported site, the popup asks you to grant host permission for that specific site (e.g. https://*.youtube.com/*). You can revoke any granted permission at any time in Settings → Sites.
| Permission | Type | Why |
|---|---|---|
activeTab |
required | Read the current tab's URL to dispatch to the right extractor |
scripting |
required | Inject the YouTube signature bridge into the MAIN world |
storage |
required | Persist settings (sync) and recent downloads (local) |
tabs |
required | Read tab metadata (title, favicon) for the popup |
contextMenus |
required | Add the right-click "GrabIt" entry on supported sites |
notifications |
required | Desktop notifications on download completion / error |
declarativeNetRequest |
required | Inject Referer/Origin headers for CDNs that require them |
declarativeNetRequestFeedback |
required | (Reserved) observe DNR rule matches for debugging |
host_permissions for each supported site |
optional | Requested on first use per-site via chrome.permissions.request() |
Why optional? If you never use GrabIt on Facebook, it never has any access to facebook.com. This minimizes the privacy surface and avoids the upfront "this extension wants to read all your data on 10 sites" warning that often scares users away.
We deliberately do not request <all_urls> or downloads (the latter isn't needed in MV3 — see AGENT_MEMORY.md for the bug history).
See the Architecture section above.
- All extension code uses ES modules (
import/export) - The service worker is declared as
"type": "module"in the manifest - Content scripts are plain IIFEs (content scripts can't easily be modules without per-file manifest entries)
- All UI text is in English; the codebase has no i18n yet (TODO)
- No external dependencies — pure Chrome Extensions API + vanilla JS/CSS
- Create
extractors/<sitename>.jsexporting an object withid,name,domains, and anasync extract(tab, opts)method returningMediaResult[] - Import and register it in
extractors/registry.js - Add host permissions and content script matches in
manifest.json - If the CDN requires referer spoofing, add an entry to
CDN_ORIGIN_MAPinlib/restriction-bypass.jsand a rule tobackground/dnr_rules.json
There's no test suite yet. Manual smoke test:
- Load the extension
- Visit each of the 8 supported sites with a typical video
- Open the popup — verify media is detected
- Click a variant — verify download starts
- Open DevTools for the service worker (
chrome://extensions→ "Inspect views: service worker`) to check logs (enable debug logging in Settings → Advanced for verbose output)
- YouTube ciphered streams — Some YouTube videos serve only ciphered URLs that require the page's JS to decode. The yt-bridge script handles many cases but not all (especially after YouTube changes the player JS).
- Age-restricted YouTube videos — Return
LOGIN_REQUIREDeven with cookies; we surface the error in the popup. - TikTok signed URL expiry — TikTok video URLs expire after ~1 hour; if you scan a tab and wait too long before clicking download, you may need to refresh.
- Bilibili video+audio split — DASH streams expose video and audio as separate variants. Downloading just the video variant yields a silent file. Muxing them into a single MP4 would require ffmpeg.wasm — planned for v1.1.
- Facebook private videos — Many Facebook videos require the user to be signed in AND be a friend/follower of the poster. We surface the "no URL found" error.
- Reddit HLS audio desync — On some Reddit posts the HLS audio variant is slightly out of sync with the video; we recommend the DASH or fallback MP4 variant in those cases.
- Instagram stories — Stories expire after 24h and use a different URL pattern than reels/posts; not currently supported.
- Twitch live streams — Live streams are not downloadable (only clips and VODs).
- Rate limiting — Aggressive scanning of many tabs in quick succession may trigger 429s from some sites (especially Reddit and Bilibili). The 30-second per-tab scan cache helps but doesn't fully mitigate this.
- No telemetry — GrabIt does not collect or transmit any data to any server other than the sites you're already browsing
- No third-party requests — All requests go directly to the sites you visit
- Cookies are sent with extraction requests so logged-in-only content works, but cookies are never read or stored by the extension
- Settings sync uses Chrome's built-in sync (encrypted, tied to your Google account)
- Recent downloads are stored locally in
chrome.storage.localand never leave your device
MIT — see LICENSE.
Built by Exon101. Inspired by projects like yt-dlp, cobalt.tools, and video-download-helper.