KeyJump

Press ; to label every link, button and field on the page, then type 1-3 characters to jump straight to it.

KeyJump lets you drive any page from the keyboard. Press ; and every link, button, form field and clickable element on screen gets a short label. Type the label to jump straight to it — buttons get clicked, links get followed, text fields get focused with the caret already at the end. Most jumps are three keystrokes or fewer. WHY IT IS FAST Labels are 1-3 characters and they are prefix-free: no label is ever the start of another one. That means the moment your typing narrows things down to a single target, KeyJump can act immediately, because there is no longer any chance you were halfway through typing a longer label. Most jumps never need Enter at all. Labels are handed out in reading order, so whatever is at the top-left of your screen — the things you reach for most — gets the single-character hints. TYPE WHAT YOU SEE If what you type does not match any label, KeyJump searches the visible text of each target instead. On a checkout page, typing "chec" filters down to the Checkout button without you ever reading a hint. The status readout tells you when this kicks in. WHY THE SEMICOLON Almost nothing else claims it. Slash and question mark are taken nearly everywhere for search and shortcut help. Period opens the web editor on GitHub and GitLab. Single letters are heavily bound by Gmail, Jira and GitLab. Semicolon is quiet, and it is on the home row. An ordinary typing character works as a trigger because KeyJump ignores it whenever your caret is in a text field. That includes code editors — Monaco and CodeMirror are both handled — so semicolon still types normally in a web IDE, even in vim mode. You can change the trigger to any single character in the options. WHAT IT FINDS Links, buttons, inputs, selects, textareas, summary elements, contenteditable regions and every interactive ARIA role. Optionally also anything that merely looks clickable, which catches the div-with-a-click-handler pattern that modern web apps are full of. It looks inside open shadow roots, checks that each target is genuinely visible rather than hidden behind a modal or scrolled out of view, and collapses nested duplicates so a button and the span inside it do not both get a hint. CONFIGURABLE - Trigger key, and the alphabet labels are built from - Light, dark, or follow the system - Hint size, and whether hints sit at the corner of an element or centred on it - Whether to include elements that only look clickable - Whether to fall back to matching visible text - Whether a single match fires immediately or waits for Enter - Per-site exclusions, so KeyJump stays out of the way where you do not want it KEYS ; show hints a-z narrow the matches Enter action the highlighted target Shift + Enter open a link in a new tab Tab / arrows step between matches Backspace undo a character Esc dismiss PRIVACY KeyJump makes no network requests. There is no analytics, no telemetry, and no server for any of it to reach. The only thing it stores is your own settings, in Chrome's own settings storage. The only permission it requests is "storage". It is open source and MIT licensed: https://github.com/vliejo/keyjump KNOWN LIMITS - Frames are hinted independently; pressing ; hints whichever frame has focus. - Closed shadow roots cannot be reached by any extension. - Chrome itself blocks extensions on chrome:// pages and on the Web Store. - Hints do not track DOM changes while they are open — press Esc and ; again. Scrolling and resizing are tracked.