Tess Browser Use

Tess browser automation extension — executes browser use commands from Tess workflows locally.

AI Agent Browser Control — Autonomous Web Navigation for AI AssistantsTurn any AI assistant into a real web operator. This extension gives AI agents secure, real-time control over your Chrome browser so they can navigate websites, read pages, click buttons, fill forms, and complete multi-step tasks on your behalf — exactly like a human would, but automated, fast, and tireless.WHAT THIS EXTENSION DOESInstead of AI models guessing what a webpage contains or relying on static scraping, this extension bridges an AI agent directly to your live Chrome session. The agent can:• Open and navigate to any URL, in the current tab or a new one • Take screenshots of the visible viewport to visually understand what's on screen • Read structured page content — text, interactive elements, links, buttons, form fields — with accurate references (refs/selectors) so it always acts on real elements, never guesses • Click buttons, links, and any interactive element • Type into text fields, search boxes, message boxes, and contenteditable areas • Scroll pages (including specific containers) to reveal off-screen content • Manage multiple tabs — open new ones, switch between them, close ones no longer needed • Wait for animations, redirects, or async content to load before proceedingAll of this happens inside a dedicated, isolated tab group, so your regular browsing tabs and personal session are never mixed with agent activity.WHY IT MATTERSTraditional automation (bots, headless browsers, scrapers) is brittle: it breaks the moment a page layout changes, it can't "see" a page like a human, and it can't reason about what to do next. By connecting an AI agent (like a large language model) to your actual browser through this extension, the agent: Sees the page as it truly renders (via screenshots), just like you do Reads real, current DOM structure and interactive elements before acting — never fabricating selectors Adapts step-by-step: observe → act → verify → observe again, the same disciplined loop a careful human operator follows Can complete long, multi-step workflows across multiple websites without human micromanagement, while still respecting login walls, CAPTCHAs, and consent screens by stopping and asking for help when blocked TYPICAL USE CASES• Research & data gathering: have an agent visit multiple sources, extract facts, and summarize findings • Form filling & data entry: automatically complete repetitive multi-field forms • Account and dashboard checks: log into internal tools/dashboards (with your existing session) and report status • E-commerce & price monitoring: navigate product pages, compare listings, extract prices • QA / workflow testing: click through a user journey and confirm each step renders as expected • Personal productivity: let an agent handle multi-click bookings, reservations, or admin tasks you'd rather delegate • Content publishing workflows: navigate CMS/admin panels to draft, review, or publish content • Social & professional networking research: browse public profiles/pages for context-gathering (respecting each site's terms of use)HOW IT WORKS (TECHNICAL OVERVIEW)The extension exposes a small, well-defined set of browser actions to the connected AI agent: Navigate: load a URL in the active or a new tab, waiting for the page to finish loading Screenshot: capture the current viewport as a PNG so the agent can visually inspect layout, text, and images Get Content: return structured page text plus an "elements" array containing every interactive element's type, ref, selector, and href — this is the ONLY way the agent identifies what to click or type into, so it never invents a CSS selector Click: trigger a click on a specific element by its ref/selector Type: enter text into an input, textarea, or contenteditable field, with an option to clear existing content first Scroll: move the viewport or a specific scrollable container by a given offset Wait: pause for a fixed short duration for animations or redirects Tabs management: list all tabs in the agent's dedicated group, open new tabs, or close tabs no longer needed This "observe → act → verify" discipline means the agent always screenshots or reads the page before deciding the next step, and re-checks after every action — minimizing errors, dead clicks, and hallucinated actions.PRIVACY & SAFETY BY DESIGN• Isolated tab group: the agent only operates inside its own dedicated group of tabs, keeping your personal browsing completely separate • No blind actions: every click or type action requires a fresh, real reference (ref/selector) obtained directly from the current page state — the agent cannot act on elements it hasn't actually observed • Respect for access barriers: if a page requires a login you haven't provided, or shows a CAPTCHA or consent wall, the agent stops immediately and reports the blocker instead of attempting to bypass it • No silent data exfiltration: the extension only relays page content and screenshots to the connected AI session you explicitly initiated — it does not run background scraping or send data anywhere else • You stay in control: you can close tabs, revoke the connection, or stop the agent's session at any timeWHO IS THIS FOR• Developers and teams building AI agents or LLM-powered automations who need real browser control, not just API calls • Power users who want to delegate repetitive, multi-step web tasks to an AI assistant • Businesses automating research, monitoring, QA, or data-entry workflows that currently require manual browsing • Anyone using an AI chat platform that offers "browser mode" and needs the companion extension to enable itPERMISSIONS EXPLAINEDThis extension requests browser tab and active-tab permissions strictly to allow the connected AI agent to view and interact with pages you navigate to during an active session. It does not track your browsing when no agent session is active, and it does not sell or share browsing data with third parties.GETTING STARTED Install the extension Open the AI platform/chat that supports agent-driven browsing Confirm the connection between the platform and this extension for your workspace Ask your AI assistant to browse, research, fill a form, or complete a task — watch it navigate in real time in its own tab group