Nextbrowser
Download
Installation4:40 watch

Run your first browser task in NextBrowser

Prepare a NextBrowser profile, connect a local agent and verify a small browser task before saving a reusable skill or custom script.

About this reference video

The recording uses OpenClaw Control to configure standalone Clawbrowser. The guide below is the NextBrowser equivalent of a first-session check, using its documented Claude Code or Codex agent connection. It does not require OpenClaw or assume a native OpenClaw integration.

Go to the NextBrowser workflow ↓
Published · @next_ossaiWatch on YouTube ↗

Overview

A first browser task should establish the connection between the intended profile, the actual running page and the agent report. NextBrowser brings those controls together in its sidebar, Chat and Live View. Start with a read-only page inspection before attempting a larger workflow.

What happens in the video

The operator pastes the installation instructions into OpenClaw Control. The agent installs the tooling, configures an API key, creates a work profile, and runs verification. Its final chat message reports successful setup and browser verification; the recording does not independently display the verification page.

Before you start

  • NextBrowser on macOS or Windows with a paired account
  • An installed and authenticated Claude Code or Codex local agent
  • A public test URL and an intended profile name

Run this in NextBrowser

Follow these steps in your own NextBrowser workspace. The video summary records the source demonstration; your result needs the checks below.

  1. Choose the profile and start its session

    In NextBrowser, select the intended profile and start the browser. Open your test URL and confirm its tab in Live View. Do not reuse another running profile merely because it is available.

  2. Select the local agent

    Choose Claude Code or Codex in NextBrowser. Confirm that the agent is installed and authenticated, then open a Chat conversation for this test.

  3. Ask for observable page facts

    Send the prompt below with the profile name and URL. Ask for title, final URL and one visible heading so the answer can be checked directly against the browser.

  4. Review before making the task reusable

    Compare the answer with Live View and page state. Once the run is reliable, you can save instructions as a skill or custom script. Its preflight should confirm the running profile, reuse or open the target tab and wait for readiness before dispatch.

Send this prompt in NextBrowser Chat

Suggested prompt

Use NextBrowser profile [profile name]. Confirm the running session, open [public test URL] and report the final URL, page title and first visible heading. Leave the page open for review. If the actual profile or page is different, stop and describe the mismatch.

Replace the placeholders, then send this in NextBrowser Chat with the intended profile and agent selected. This is a suggested task, not the recording transcript.

What to verify in your run

  • A first-task report that matches the page visible in NextBrowser
  • A verified starting context that can be reused in a skill or custom script

Questions about this workflow

Do I need OpenClaw for this guide?

No. The video uses OpenClaw, but these instructions use the documented NextBrowser local-agent workflow with Claude Code or Codex.

Does selecting a profile constrain every agent action?

No. NextBrowser passes the selected profile as context. Confirm the actual profile and session in the sidebar and Live View before accepting the result.

Keep exploring

NextBrowser documentation

NextBrowser documentation ↗Product guide: profiles, agents and sessions ↗

From the blog

Self-Hosted Browser Automation: The Complete Guide for 2026 ↗

Read about preparing and operating a browser automation environment.

Related use cases

AI Agent Browser Automation ↗

NextBrowser workflow adapted . Recording reviewed . Video source: published video and description.

More workflows to explore

← Browse all demos