A free, local, watermark-free Chrome extension (Manifest V3) for screenshots and screen recording. Built because being nagged to log in just to save a screenshot is miserable.
- Free & unrestricted — no paywalls, no watermarks, no sign-ups, no upsell.
- Private & local — every byte of capture, processing, and storage stays on your machine. No servers, no telemetry, no cloud, no runtime network requests (fonts are vendored, not fetched).
- No dark patterns — no forced new tab after recording, no mandatory cloud hosting. Capture → it lands in your Downloads. Done.
Screenshots (PNG)
- Visible Tab · Full Page (auto-scroll + stitch, de-dupes sticky headers) · Select Area (on-page green marquee with live dimensions, Enter to capture)
- After capture you get a captured card: Annotate & save, Save PNG directly, Copy, or Discard
- Annotation editor — Draw, Arrow, Rectangle, Text, Highlight, Blur/Redact, Eraser, Crop (free-form or 16:9 / 1:1 / 9:16 / 4:3), a Layers panel (select, reorder, show/hide, delete, re-edit text, add image/logo layers), a Beautify backdrop (gradient/solid + rounded corners + shadow), colours, stroke weights, undo/redo, copy-to-clipboard
Screen recording → native .mp4
- Current Tab — instant, no picker; a Loom-style control bar sits on the page (auto-hidden during capture)
- Screen / Window — Chrome's native picker (
getDisplayMedia, run inside the offscreen document): record a whole monitor or any app window, even outside Chrome - Screen + Cam — the picker plus your webcam composited as a corner bubble (circle/square, size, corner, mirror — all adjustable live), drawn onto a canvas in the offscreen document and recorded as one MP4
- Screen / Screen+Cam are controlled from the toolbar REC badge, the popup, and keyboard shortcuts (no on-page bar off-tab, and no countdown — the picker is the "get ready" beat)
- System / tab audio and/or microphone (mixed), pause / resume, and a red REC badge
Video editor (opens after a recording; WebCodecs, no ffmpeg/WASM)
- A multi-track timeline — pinned ruler + video track, then a track per overlay, plus audio — with drag-to-move / drag-edge-to-resize clips, snapping (playhead, clip edges, whole seconds), timeline zoom, and a constant-height panel (adding tracks never reflows the stage)
- Trim the ends and cut mistakes out of the middle (multi-segment), resolution and speed
- Crop free-form or to a ratio (16:9 / 1:1 / 9:16 / 4:3); Zoom with draggable, resizable focus blocks that ease in/out; Beautify backdrop (gradient/solid + rounded corners + shadow)
- Layers — annotations (text, arrows, shapes, blur), image/logo overlays, and animated GIF overlays — all resizable on the stage (8-handle selection box) and time-scoped on the timeline (e.g. an arrow that appears for 3 seconds), composited live (WYSIWYG preview = export)
- Audio editing — waveform view, drag to mute regions (without cutting the video), volume / mute, and import a voiceover or music track (movable, trimmable, mixed into the export)
- A left tool rail with keyboard shortcuts (V R A T B I), draw-once tools that return to Select, ⌘D duplicate, frame-step keys, and prev/next-edit transport buttons
- Export a native MP4 (bitrate scales with resolution, so 720p/1080p genuinely shrink files) or an animated GIF — or download the original clip untouched
Google Drive backup (optional, off by default)
- Connect your own Google Drive in the popup and screensnap can auto-upload finished recordings to a private "screensnap" folder — or upload one-off from the done card / the editor's export menu
- Direct browser-to-Google only (narrow
drive.filescope: screensnap can only see files it created). No screensnap servers, no proxy, no account with us — disconnect revokes access. Until you connect it, the extension makes no network requests at all. Maintainer setup: docs/DRIVE_SETUP.md
MediaRecorder records H.264 MP4 directly on modern Chrome, so there's no ffmpeg, no WASM, no
conversion step — the recording is saved the instant you stop. (On the rare build without native
MP4 support it saves .webm rather than lose your recording.)
Most people: add it from the Chrome Web Store — one click, no sign-up.
From source (for development):
chrome://extensions → enable Developer mode → Load unpacked → select this folder
That's it — no build step, no npm install, zero dependencies. Icons and fonts are committed, so
it runs straight from source. (npm run make:icons only re-generates the placeholder icons.)
No framework, no bundler, zero runtime dependencies — plain ES modules + self-contained injected overlays.
| Piece | File(s) | Role |
|---|---|---|
| Service worker | src/background/service-worker.js |
Orchestrates everything: screenshots (captureVisibleTab + injected scroll/crop helpers stitched with OffscreenCanvas), the recording lifecycle, downloads, badge, and persisted state. |
| Popup | src/popup/ |
The screensnap UI — capture/record tabs, captured card, recording / saving / done. UI + dispatch only. |
| Offscreen document | src/offscreen/ |
Hosts MediaRecorder (a service worker has no DOM; the popup closes on blur). Opens the screen picker (getDisplayMedia), and for Screen + Cam composites the screen + webcam bubble onto a canvas (driven by draw-worker.js). Saves native MP4. |
| On-page overlays | src/content/ |
Injected via executeScript: editor-overlay (annotation editor), recorder-control (the recording control bar: countdown → timer + pause + discard + stop) and draw-overlay (auto-fading pen). Re-injected on navigation so controls survive reloads / navigation. The Screen + Cam webcam is not an overlay — it's composited into the video in the offscreen document (draw-worker.js metronome + canvas), so it works over any screen/window. |
chrome.tabCapture records the entire tab, and there is no API to exclude an on-page element — so a visible
control would appear in the video. Like Loom, the control bar copies that constraint instead of fighting it: it's
bottom-left and collapsed / auto-hidden during recording (move the pointer to the bottom-left corner to reveal
it), and global shortcuts (commands: stop Alt+Shift+S, pause Alt+Shift+P) + the toolbar REC badge stop/pause it
while hidden — so the recording stays clean. (Screen / Screen+Cam capture at the OS level via getDisplayMedia, so
their controls are the badge / popup / shortcuts; the on-page bar there is a best-effort extra on the active tab.)
Surviving a jump to a different site mid-recording needs the optional <all_urls> host permission, which the
popup requests with the user's click when a Current Tab recording starts. The captured media stream keeps running
across navigations regardless.
State lives in chrome.storage.session so it survives the service worker being torn down
mid-recording. The popup, recorder window, and overlays all reflect it live via STATE_CHANGED.
activeTab + scripting (only the tab you invoked it on — no <all_urls>, no history scope),
tabCapture (record the current tab), offscreen, downloads, storage. Whole-screen / window capture
uses the standard getDisplayMedia() Web API from the offscreen document — it needs no extra
permission and shows nothing until you start a Screen / Window recording and pick a source (no background
screen access). Screen + Cam additionally uses your webcam via standard getUserMedia (one-time browser
prompt) — also no manifest permission. For the optional Google Drive backup, identity ships in the
manifest (it is warning-free and inert until you connect — Chrome can't reliably expose a
runtime-granted API to a running service worker, so it can't be optional), while googleapis.com
host access is optional and requested only when you click Connect.
- Full-page capture uses
captureVisibleTab(rate-limited ~2/sec), so long pages take a few seconds; lazy-loaded content and complex sticky layouts can still leave seams. - Screen + Cam composites the webcam as a fixed corner bubble (pick corner/size/shape/mirror in the popup, adjustable mid-recording) — it isn't a freely-draggable live overlay, because off-tab screen capture has no web page to overlay.
- Screen / Window capture has no on-page overlays when you're off the active browser tab (the camera bubble and pen live in a web page), and on macOS system audio isn't captured (no OS loopback) — the mic still records. The native picker / OS prompt governs screen access.
- Internal pages (
chrome://, the Web Store, etc.) can't be captured as a tab — Chrome blocks it. (You can still capture them via Screen / Window, which records at the OS level.)
UI type is Geist / Geist Mono (SIL OFL 1.1), vendored as
.woff2 under src/popup/fonts/ so nothing is fetched at runtime.
See CLAUDE.md — trunk-based commits to main, zero runtime dependencies, and the
privacy/free principles above are non-negotiable.
GNU GPL v3.0 (or later) — see LICENSE.