xbrowser

What problem does it solve?

It removes the need for brittle, terminal-less browser scripting by enforcing a single, reliable automation path through xb for any real web interaction.

Core Features & Use Cases

• Exclusive browser automation: All browser tasks (open, click, fill, screenshot, scrape, navigate, and web-app testing) must go through xb instead of agent built-ins.
• Session and login-state reuse: Reuses real browser profiles via CDP-based automation to avoid repeated logins.
• User-decision guardrails: When xb returns awaits_user_input, the skill surfaces the UI options to the user and executes only the mapped command after user selection.
• Batch and deterministic execution: Supports batch sequences for deterministic steps while preserving JSON-by-JSON error checking.
• Failure recovery: Provides guidance for timeouts, @ref invalidation recovery (re-snapshot), and headed debugging.

Quick Start

Use the xbrowser skill to open a page and capture a snapshot of interactive elements in one guided flow: open https://example.com.