Capture. Extract. Act.
Smart screenshot tool that captures, runs local OCR, classifies the content, and suggests the right action. Four capture modes, eight collections, zero servers.
Four ways to capture
Every screenshot Chrome can take. None it can't.
Visible tab
Snap what's on screen right now. One keystroke, one shot.
Area select
Drag a rectangle. Pixel-precise crop, snaps to your cursor.
Full page
Scroll-stitched capture of the entire page — header to footer.
Screen / window
Native desktop picker. Capture any app, any monitor.
What's inside
A screenshot tool that thinks for you
Local OCR via Tesseract WASM
Text extracted on-device in an offscreen document. Images are preprocessed and tiled vertically for tall captures. No cloud round-trip, works offline.
Auto content detection
Confidence-scored tagging into code, prices, contacts, tasks, UI, plain text, or mixed — so the right action shows up first.
Context-aware actions
Copy code, save lead, set reminder, create task — actions adapt to what you captured and your active mode.
Search OCR + metadata
Find any past screenshot by the words inside the image, the page title, the URL, or your tags.
Collections out of the box
Code Snippets, UI Inspiration, Prices to Track, Leads, Tasks, Reading Notes, Job Opportunities, Personal Archive — all preseeded.
Side-panel library
Grid view, multi-select, type filters, viewer with action bar. Everything stored locally in IndexedDB.
Keyboard-first
⌘⇧S for visible tab, ⌘⇧A for area. Right-click anywhere for context-menu capture options.
Smart tagging
Regex detectors plus content-type heuristics auto-tag each capture — so search is instant, even months later.
Resume-on-startup
If a long OCR job is interrupted, ScreenMind picks it back up on next launch. No half-processed captures left behind.
The pipeline
Capture → Extract → Act. Every time.
Capture
Pick a mode — visible tab, area, full page, or screen. Hotkey or right-click, your choice.
Extract
Tesseract runs in an offscreen document. Image preprocessed for contrast, tiled for tall pages, OCR'd locally.
Act
Content type is detected. The right action — copy code, save lead, create task — surfaces first. Take it or file it.
Suggested actions
The right next step, already queued up
Snap a code block — "Copy code" is first. Snap a contact card — "Save lead" is first. Switch modes and the same captures get a different ranked menu. No menu-hunting.
- Auto-detects code, prices, contacts, tasks, UI, plain text
- Confidence-scored tagging powers search and filters
- Mode-based re-ranking (Dev / Business / Shopper / Research)
- Right-click context menu mirrors the side-panel actions
ScreenMindSuggested actions
- Copy code⏎
- Save to Code Snippets
- Create task
- Add tag
Modes
Tune ScreenMind to what you actually do
Dev
Boosts Copy Code and Create Task. Default mode for engineers.
Business
Boosts Save Lead and Create Task. Contact cards become leads in one click.
Shopper
Boosts Copy Text and Create Task. Price captures get tracked instantly.
Research
Boosts Copy Text and Add to Collection. Reading notes pile up clean.
Your screenshots never leave your machine.
OCR runs locally via Tesseract WASM in an offscreen document. The extension makes zero outbound requests — verified by grep across the source. No analytics, no auth, no remote config. Captures live in your browser's IndexedDB and are deletable from the side panel in two clicks.
Keyboard
Two keys, two captures
Or right-click → ScreenMind for the context menu
Questions
The honest answers
No. OCR runs locally via Tesseract WASM inside an offscreen document. There are no fetch calls to external hosts in the extension — only chrome.runtime.getURL for bundled assets. No telemetry, no analytics, no remote config. Audit the source to verify.
In your browser's IndexedDB via Dexie. Image blobs, thumbnails, OCR text, tags, and collections — all on your machine. unlimitedStorage is requested because screenshots are bulky. Clear the library from the side panel any time.
Tesseract.js v7 with English. Preprocessing (grayscale, contrast, binarize) plus vertical tiling lifts accuracy on long pages and small UI fonts. It's not a cloud OCR, so screenshots of stylized fonts or rotated text are weaker. For most code, prices, and contact info it's solid.
Modes re-rank the suggested actions. Dev surfaces 'Copy Code' first, Business surfaces 'Save Lead', Shopper surfaces price actions, Research surfaces collection adds. Same captures, different first-action.
Yes — host_permissions: <all_urls> means capture works on any page you visit. Chrome's own internal pages (chrome://) are sandboxed for everyone and can't be screenshotted by any extension.
Free. MIT licensed. Source on GitHub. No accounts, no paid tier.
Free · MIT · v0.3.0
Stop hunting through screenshots. Start using them.
One click from the Chrome Web Store. No account, no setup, no upsell.