Note: This skill overlaps with browser-use. Prefer agent-browser for Node.js/rust-centric workflows or direct CDP control; prefer browser-use for Python-centric workflows. Tool access belongs in the agent client's permission config; this skill documents workflow only.

Browser Automation with agent-browser

The CLI uses Chrome/Chromium via CDP directly. Install via npm i -g agent-browser, brew install agent-browser, or cargo install agent-browser. Run agent-browser install to download Chrome.

Core Workflow

Every browser automation follows this pattern:

1. Navigate: agent-browser open <url>
2. Snapshot: agent-browser snapshot -i (get element refs like @e1, @e2)
3. Interact: Use refs to click, fill, select
4. Re-snapshot: After navigation or DOM changes, get fresh refs