What problem does it solve?
Debugging or interacting with web pages often requires inspecting real browser state (DOM, console/network events, and screenshots) that can’t be reliably reproduced from text alone.
Core Features & Use Cases
- Inspect page structure with semantic snapshots: Generate a compact, agent-friendly view of visible interactive elements for navigation and reasoning.
- Extract HTML and evaluate expressions: Pull full page HTML or targeted selector HTML, and run safe JavaScript expressions against a specific browsing context.
- Drive interaction with reliable clicks and typing: Click by CSS selector or accessible text, click at coordinates as a visual fallback, type into the focused element, and repeatedly load more content.
- Diagnose console and network issues: Subscribe to BiDi console events and network responses (and optionally use experimental RDP for closer DevTools Console parity).
- Manage a Firefox instance: Reuse a managed Firefox profile/process across commands, or connect to an existing BiDi-enabled Firefox using environment configuration.
Quick Start
Run the skill to list contexts and auto-start a managed Firefox if needed, then choose a context prefix for subsequent inspect and interaction commands.