| name | Browser Automation |
| slug | browser-automation |
| description | Load when the user wants to navigate a website, click through a flow, fill a
form, extract data from a page, or record an interaction as a reusable
playbook.
|
| icon | Globe |
| color | #ef4444 |
| version | 2.1.0 |
| category | automation |
| tools | ["browser_open","browser_click","browser_type","browser_screenshot","browser_extract_text","browser_extract_data","browser_wait","browser_select","browser_scroll","browser_navigate","browser_close","browser_record_start","browser_record_stop","browser_record_status","list_playbooks","load_playbook"] |
| config_schema | {"type":"object","properties":{"default_timeout":{"type":"integer","default":30000}}} |
Default operation timeout: {default_timeout}ms.
Always read the interactive-elements summary returned by browser_open before clicking — guessing selectors wastes turns.
CSS selectors are more reliable than text matching. Use browser_wait before interacting with anything that loads dynamically.
For repeat workflows: check list_playbooks first. To capture a new one, browser_record_start → do the task → browser_record_stop saves it as .md.
When you hit a CAPTCHA, popup, or unexpected state, take a screenshot and ask the user — don't click around hoping to escape.
Always browser_close when done; sessions hold significant resources.
See references/selectors.md for selector cookbook (login forms, infinite scroll, modal dismissal).
Gotchas
