t·
Handle
your whole Chrome, readable by your agent
your chrome, readable — and drivable — by your agent

Your whole Chrome, one question away.

Your AI reads the signed-in tabs you have open and everything behind them — your history, downloads, the pages you closed — searched by meaning, not keywords, in plain language. Ask it "what did I work on today?" or "pull up that article I read last week" and it actually can. And when you ask, it can click, type, and watch a tab too — even behind a strict CSP.

Works as an MCP server (25 tools), a CLI, and a board. Local, no API key, open source.

Your browser knows everything. Nothing can use it.

Your browser is the highest-fidelity record of your work and attention — every doc, dashboard, article and design you've touched. Yet your AI can't see any of it, and Chrome itself can barely search it.

Today — locked away
  • Your AI can't read the page in front of you — you copy-paste it in, one tab at a time.
  • It's blind to anything behind a login — your docs, dashboards, work tools.
  • Chrome's history is a keyword-and-time list — no searching by what a page was about.
  • The article you read last week, the tab you closed an hour ago — gone unless you remember the title.
  • Nothing can answer "what was I working on yesterday?"
With Handle
  • Your AI reads what's on your screen — just say "use the doc I'm looking at."
  • It reads your signed-in pages — the Google Doc, the dashboard, the SharePoint file.
  • It searches your history by meaning — "that pricing article I read last week."
  • It reopens what you closed and pulls back the page you forgot the name of.
  • Ask "what did I work on today?" — and it tells you.

Just say it — examples

You talk to your AI in plain language; it runs the right Handle tools for you. A few things people ask:

> what did I work on today?
Reads your history back as themes, not a raw list.
> pull up that Stratechery piece I read last week
Finds it in your history by meaning → reopens it → summarizes.
> summarize the Google Doc I'm looking at
Reads your signed-in page, past the login wall.
> answer this from the six tabs I have open
Reads all six and answers cited by tab.
> reopen the doc I closed this morning
Recovers it from the session and opens it back up.
> compare these three pricing pages
Reads each and lays out the differences.
> read the PDF I just downloaded
Resolves the download to the local file and reads it.
> what does this Figma show?
Screenshots the board and reads it with vision.
> what errors are on this page?
Pulls the console logs (with the extension).

Under the hood: it reads your real browser

Most "AI that browses" tools open a fresh, robot-controlled Chrome signed into nothing — a stranger to your work. Handle reads the browser you're already logged into. Programs are walled off from the browser by design, so it keeps several ways in — a Chrome add-on, the built-in macOS bridge, or an opt-in developer channel — and tries the smoothest first. Because there's always a backup, a read almost never just fails.

Other tools — the raw page~50,000 words of code
Handle — just the content, tidied up~800

Handle hands your AI only the part you'd actually read — clean Markdown, not the page's markup and scripts. Roughly 95%+ less to chew on: cheaper, faster, and less for the model to get lost in.

It always gets you something

Behind one plain "read the tab," Handle tries the cleanest route first and drops to the next only if it must — so you get usable content, or a clear reason why not.

  1. A snapshot it already hasCaptured a moment ago — instant, nothing to re-fetch.
  2. The live page, content onlyJust the article — headings, lists, tables — as clean Markdown; menus and ads dropped.
  3. The document's own copyGoogle Docs & Sheets, and Word on SharePoint, pulled from your signed-in session.
  4. A picture, read with visionDesigns, PDFs, anything with no text — screenshotted, and your AI reads the image.
  5. Your clipboardLast resort — you copy, it grabs it. There's always a way through.

The tools your agent gets

Handle exposes 31 tools over MCP, so any agent can use them as first-class functions. You rarely call these by name — your AI picks the right one — but here's the full surface.

Read & reference

read_tabpage text — HTML, Google/Office export, or screenshot, by type active_tabthe tab you're looking at right now find_tabresolve "the figma one" to a tab grep_tabssearch the on-page content of every open tab list_tabsevery open tab, with groups + stale flags crawlfollow a collection's links → each page's full text, combined screenshot_tabcapture a tab as image(s) to read with vision grab_clipboardread what you copied — the reliable floor

Ask & keep research

ask_tabsanswer a question across your open tabs, cited search_webfull-text search every page you've read through Handle save_tabscapture tabs into a dated research bundle recall_bundleload a saved bundle back as context list_bundleslist your saved research

Your browser's past

historyquery your history by meaning (+ search terms) closedrecently-closed tabs, real titles → reopen bookmarkssearch your bookmarks downloadsrecent downloads → local file paths journeysChrome's own topical clusters of your browsing most_visitedthe sites you live in

Act on tabs

click_tabclick an element — by selector, text, or intent (CSP-safe) type_tabtype into a field, optionally submit (CSP-safe) meetingscrape a Zoom recording's transcript → JSON + Markdown open_tabfocus a tab, or open any URL close_tabclose tabs for real group_tabmove a tab into a group note_tabattach a note to a tab pin_tabpin / unpin a tab refresh_tabsre-scan Chrome into Handle

Sturdier backend

consoleconsole logs + errors for a tab (via the extension) ext_statusis the Chrome extension connected auditthe append-only trail of what Handle did in your session read_tab --cdpread via DevTools Protocol when you've opted in

Connect it in one line: claude mcp add handle -- node ~/claude-apps/handle/mcp/server.js

Set it up once

You'll need macOS, Google Chrome, an AI assistant you already use (like Claude Code), and Node + Python 3 for the local bridge (most dev machines already have both). It's free and open source, runs entirely on your own computer — no account, no API key. It reads your tabs; it never sends them anywhere, and it can't touch anyone else's.

1

Get it

git clone https://github.com/abhitsian/handle.git ~/claude-apps/handle
2

Connect it to your AI

One line, and your assistant can read your tabs:

claude mcp add handle -- node ~/claude-apps/handle/mcp/server.js
3

Let it read your pages — two one-time settings

a. The first read pops a macOS prompt to let it control Chrome — click OK.
b. In Chrome, turn on View → Developer → Allow JavaScript from Apple Events. This one's a menu toggle, not a popup, and it's off by default — it's what lets Handle read the actual text on a page. Skip it and tabs list fine but reads come back empty.
c. Only for screenshots of designs and PDFs (like "what does this Figma show?"): grant Screen Recording to your terminal in System Settings → Privacy & Security. Text reads don't need it.

4

Just talk to it

"Summarize the doc I'm looking at."   "What does the figma show?"   "Answer this from my open tabs."   That's the whole thing.

Optional: there's a small Chrome companion that makes it sturdier — your real tab groups, cleaner screenshots, and page errors — installed once, no browser restart, and it talks only to your own computer. Most people never need it; the steps above work on their own.