The web, adapted to you.
43 tools across vision, reading, motor, cognitive, and the everyday annoyances, plus read aloud and a guide that points at the control you need next. Press Ctrl+Shift+U (⌘+Shift+U on a Mac) to open the panel on any page. Every tool here is free for individuals, with no account. The AI tools additionally need either your own Google Gemini key or a subscription, because somebody has to pay the model.
3 tools active on example.com
See it in action
One keypress changes everything.
Reader mode rebuilds the article on the page you already have open. No reload, no second request, nothing sent anywhere.
Scientists discover ancient Roman shipwreck off the coast of Sicily
By Jane Doe · Nov 12 · 4 min read · Filed under: History, Science
The wreck, discovered at a depth of 80 metres by a team from the University of Palermo, dates to approximately 200 BCE and contains amphorae still sealed with their original contents. Initial analysis suggests the cargo included olive oil, wine, and garum.
"This is one of the most significant finds in Mediterranean archaeology in the last decade," said lead archaeologist Prof. Marco Russo.
A typical article page: ads, banners, sidebar, related content.
Vision
For low vision, color vision differences, and anyone who needs the page to look different.
Dark mode
Turns a light site dark with tuned colors rather than a blanket invert, so photos and logos survive.
High contrast
Pushes contrast until faint grey-on-white body text becomes readable. Three strengths.
Color blindness filter
Deuteranopia, protanopia, and tritanopia presets, applied as a CSS filter matrix. No reload.
Text size
Scales text in steps without scaling the browser UI. The layout reflows; nothing is clipped.
Invert / dark
A plain photographic negative for sites dark mode cannot reach. Images are flipped back the right way.
Saturation
Full grayscale, or dial color intensity up and down for sensory comfort.
Custom colors
Pick your own text, heading, and background colors when none of the presets is the one you need.
Magnifier
A lens that follows your pointer and enlarges whatever is under it, without zooming the page.
Reading
For dyslexia, focus, and anyone who reads better with a different shape.
Reader mode
Extracts the article and reformats it into a clean, centered column. Runs on the page you already loaded.
Dyslexia font
OpenDyslexic, Lexend, or Atkinson Hyperlegible across body text, headings, inputs, and buttons.
Readable font
Replaces every typeface on the page with one clean, legible sans-serif.
Bionic reading
Bolds the opening letters of each word so your eye has a fixed point to land on.
Text spacing
Line height, letter spacing, and word spacing move independently, so you can loosen only what is tight.
Text alignment
Forces one consistent alignment on body text, including away from justified.
Highlight links
Underlines and outlines every link so they stop hiding inside sentences.
Highlight titles
Boxes and labels each heading with its level, so the page outline becomes visible.
Reading guide
A ruler line that tracks your cursor to hold your place in a paragraph.
Reading mask
Dims the page except a band that follows your cursor.
Focus mode
Dims everything except the paragraph you are reading, and follows you down the page. Reading position, not cursor position, so it works without a mouse.
Motor
For trembling hands, reduced fine motor control, and anyone who needs bigger targets.
Bigger click targets
Pads buttons, links, and inputs out to a comfortable minimum tap size. They stay where they look.
Big cursor
A large, high-contrast pointer that is easy to keep track of.
Visible focus
Forces a high-contrast ring on whatever has keyboard focus, on sites that removed theirs.
Auto-scroll
Scrolls the page slowly and hands-free at a speed you set. Moving the mouse pauses it.
Force zoom
Re-enables pinch-to-zoom on sites that disabled it in their viewport tag.
Pin tooltips
Hover or tab to anything with a tooltip and it stays open until you dismiss it, instead of vanishing the moment your hand moves.
Unpin sticky bars
Unpins headers that follow you down the page, and hides floating chat bubbles.
Cognitive
For ADHD, sensory sensitivity, and anyone who needs the noise turned down.
Hide distractions
Collapses ads, related-content sidebars, and newsletter clutter, by common markup patterns.
Stop animations
Freezes CSS animation and transitions, and stops autoplaying video.
Hide images
Blanks images and background art for focus or low bandwidth. Text stays intact.
Mute media
Silences audio and video, and stops autoplay from un-muting itself.
Feed off (stop the scroll)
Hides the infinite feed on YouTube, Reddit, TikTok, Facebook, and LinkedIn. Search and subscriptions still work.
Mute notification nags
Hides "turn on alerts" bars, and answers browser notification prompts as no before they interrupt you.
Annoyance killers
For everyone tired of the modern web.
Cookie banners
Finds the consent banner and hides it. Clicking reject for you is a separate switch, off by default, because a consent choice is yours to make.
Dismiss popups
Hides newsletter and promo overlays and gives you the scrollbar back.
Break sign-up walls
Peels off "sign up to keep reading" overlays and un-blurs the article underneath.
Expand all
Opens the show-more and FAQ blocks inside the article, once, when you switch it on. It stays out of navigation, forms, and checkout.
Unlock selection
Re-enables selecting text everywhere, and defeats most copy and right-click blockers. Editors and form fields keep their own behaviour, so Docs still works.
Media speed
Sets playback speed for every video and audio element on the page.
Volume boost
Makes quiet video louder and tames spikes, without touching system volume. Media the browser will not let it process safely is left alone rather than silenced.
Picture-in-picture
Pops the page video into a floating, always-on-top window.
AI reading aids
Three tools that call a model, with your own Gemini key. Off until you turn them on, and each one tells you what it sends.
Highlight to explain
Needs a keySelect text for a bubble that explains, simplifies, or translates it. Define and Speak in the same bubble need no key.
Double-click dictionary
Double-click a word for a definition and a button to hear it. Uses the free dictionaryapi.dev, no key and no model.
TL;DR this page
Needs a keyPins a short summary above the article. Turning it on sends the article text to Google’s Gemini API.
And 7 that are not toggles
7 toolsRead aloud
Reads the selection, or finds the article text when nothing is selected. Web Speech API, so the audio is synthesised by your browser and no text leaves the device.
Guide
Needs a keyType or speak what you are trying to do and it answers as ordered steps, pointing a cursor at the exact control to use next. Grounded in the page’s own elements, so the pointer tracks scroll and reflow. Needs a Gemini key.
Fix this site
Point at anything that gets in your way and hide, collapse, widen, or restyle it. The fix is CSS and DOM only, saved for that site, and re-applied next visit.
Clean copy
Copies selected text without the tracking parameters, invisible characters, and "read more at" tails sites append to your clipboard.
Ask this page
Needs a keyType a question about the page you are on and get the answer in the panel. It reads the article text to answer; it does not scroll the page or highlight anything for you. Needs a Gemini key.
Read the text in images
Needs a keyPulls the words out of screenshots, infographics, and image-only documents, up to four at a time. Needs a Gemini key.
Describe images
Needs a keyOne button writes alt text for the images on the page that have none, up to six at a time, using the page title as context. Images the browser will not let it read are skipped. Needs a Gemini key.
50
tools in the panel
6
categories
6
need an AI key
∞
free for individuals
What leaves your device
The exact answer, not the comfortable one.
Most of this extension never talks to a server. The AI tools do, because they cannot work otherwise, and this is what they send and to whom. There is no analytics and no telemetry: we never collect usage data, and nothing is sent to us for any purpose other than answering the request you just made. An account is optional and exists for one reason, which is that the AI tools otherwise require you to obtain a Google Cloud API key yourself. Bring your own key and the AI still works and still never touches our servers.
Never leaves your browser
- Everything under Vision, Reading, Motor, Cognitive, and Annoyance killers. They are CSS and DOM changes made in your browser.
- Read aloud. Your browser synthesises the speech with the Web Speech API.
- Your per-site tool settings, your "fix this site" rules, and your friction notes, including any screenshot you attach. They are saved on this machine and go nowhere.
Follows your Chrome profile
- The panel theme and your chosen read-aloud voice, through Chrome sync, so they follow your signed-in profile. Nothing about which sites you visit goes with them.
Leaves your browser, when you use these tools
Google (generativelanguage.googleapis.com)
Sends: The text you selected, or the article text, or the images being described or read, or a numbered list of the page’s elements for Guide. The page title and URL go with a summary, a question, or a Guide answer so the answer makes sense.
When: Only when you use Highlight to explain, TL;DR this page, Ask this page, Describe images, Read the text in images, or Guide, AND you are using your own Gemini key rather than a subscription. With your own key the request goes straight from your browser to Google, and Google’s terms for that key apply to you.
WebAble (api.webable.cc)
Sends: Exactly the same content as the row above: the selection, the article text, the image, or the numbered element list, plus the page title and URL where they are needed. It is passed straight through to Google and the answer is passed back.
When: Only when you use one of those AI tools AND you are signed in to a subscription instead of using your own key. This is the whole difference a subscription buys: we hold the Google key so you do not have to.
Google speech (through your browser)
Sends: The audio recorded while you hold the Guide’s microphone button. Chrome’s built-in speech recognition transcribes it on Google’s servers, and asks the site you are on for microphone permission.
When: Only while you are pressing that button. Typing your question instead sends no audio.
api.dictionaryapi.dev
Sends: The single word you double-clicked, or the first word of your selection. Nothing else, no page URL, no identifier.
When: When Double-click dictionary or the Define button is used. No key and no account.
Whatever host serves an image on the page
Sends: A fresh request for that image file, so it can be read and sent on to Google. No page URL, no identifier; the site is already serving that image to you.
When: When Describe images or Read the text in images runs.
And what is kept
Our database
A count of how many AI requests you made today, and which kind the last one was. Verified in the code: no request body, no page text and no URL is written to any table.
Our logs
Nothing from the request body. The server logs errors, not content. If Google returns an error, Google’s error message is logged, and we cannot rule out that such a message quotes a fragment of what was sent.
Google
Not ours to promise. Their retention depends on the terms that apply to the key in use, and with a subscription that is our key rather than yours. We have not published a specific claim about it because we have not verified one.
Your Gemini key is stored on the device you typed it on and is never synced, so it does not follow your Chrome profile to your other machines. That means pasting it again on each machine you use. Anyone with access to your computer can read it, as with any key held in a browser, so on a shared machine use a hosted proxy or skip the AI tools entirely.
Free to install · no account required
Be first to install it.
Chrome first. We'll email once when it ships. No newsletters, no drip campaigns, no upsells.
Free to install · Account optional · No telemetry