Name: Agent Browser
Author: TheSethRose

What breaks without openclaw agent browser skill

Agents stalled at first click. Manual link-following. Research that never completes.

→

Autonomous multi-hop browsing × 185-star agent-native skill ÷ 15–25 minutes ÷ no manual navigation = deep research delivered hands-free.

openclaw agent browser skill — what it actually does

Lets agents autonomously navigate websites and follow links to a goal.

Extracts structured data from pages visited during autonomous sessions.

Combines with note-taking skills to build full research pipelines.

Sets MAX_HOPS to prevent runaway sessions on large sites.

Security check — openclaw agent browser skill

Privacy score: 7/10 — accesses connected platform APIs only. Lock it: review OAuth scopes before install, confirm macOS, Linux; OpenClaw ≥1.0; Node.js ≥18 compatibility.

Quick start — openclaw agent browser skill in 15–25 minutes

Setup time: 15–25 minutes

You need:

OpenClaw core
Playwright or Puppeteer

Install the package:

# Install via ClawhHub
clawhub install thesethrose/agent-browser

Install browser dependencies: npm install puppeteer

Install the skill

Set AGENT_BROWSER_ENGINE=puppeteer in .env

Run /agent-browse <goal> to start autonomous browsing

Troubleshooting openclaw agent browser skill

1. Autonomous mode can follow many links — set MAX_HOPS to avoid runaway sessions

2. Login-protected pages require manual cookie injection

Compatibility & status

Works with: macOS, Linux; OpenClaw ≥1.0; Node.js ≥18 intermediate Last updated: Nov 2025 ★ 185 on GitHub MIT

Official docs →

View on GitHub →

Related — more like openclaw agent browser skill

More by TheSethRose

Keep babysitting every research step and watch hours disappear. Deploy autonomous browsing before the next deep-research task.

Get it on GitHub →