Docs

Getting started.

Install Limitless Browser, sign in, and run your first agent task. The whole first run takes about two minutes, with no API keys or models to configure.

What Limitless Browser is

Limitless Browser is the first desktop super app: an AI agent that operates a real browser using 130+ built-in tools. You describe a goal in plain language and the agent plans the steps, executes them across tabs, pauses to ask your approval before anything sensitive, then reports back what it did. It keeps a persistent, inspectable memory so it gets more useful the more you work with it.

The AI service is included. There is nothing to wire up: no API keys, no model selection, no separate backend. Sign in and start typing tasks.

1. Install and open the app

Grab the build for your platform from the download page and follow the installation guide for macOS, Windows, or Linux. Launch it the same way you would any other desktop app.

2. Sign in

On first launch you'll be prompted to create an account or sign in. Your account carries your plan and credit balance and anchors the agent's memory. New accounts start on the Free plan, which includes a monthly allotment of credits, enough to run real tasks before deciding on Plus or Pro.

3. Learn the interface

The window has four parts you'll use constantly:

  • Command bar. where you type a task in plain language. This is the main way you talk to the agent.
  • Live view. the real browser the agent is driving. You watch every navigation, click, and keystroke as it happens.
  • Plan and activity panel. the agent's step-by-step plan and a running log of what it has done.
  • Approval prompts. inline requests that appear before any sensitive action, so you stay in control.

4. Run your first prompt

Focus the command bar with Cmd/Ctrl + K and describe something concrete. Here is what a run looks like: the agent shows its plan, executes it in the live view, and pauses at any gate for your approval.

limitless://agent · hacker-news
Go to Hacker News and list the top 5 stories about AI posted today, with links.

Plan

Open Hacker News
Scan today's front page28 stories read
Filter to AI topics and rank
Return the top 5 with links
APPROVAL REQUIREDSend the summary to the team channel
ApproveDeny

Approve any prompts that appear along the way. For a detailed, step-by-step walkthrough, continue to Run your first task.

Where to go next

  • First task: write good prompts and read the agent's results.
  • Permissions: approval gates, take-over, and per-site rules.
  • Memory: what the agent remembers and how to inspect it.
  • Troubleshooting: fixes for common issues.