What problem does it solve? Automating browser tasks normally requires separate automation frameworks with fresh, logged-out sessions. This Skill drives the user's actual browser through a local daemon at 127.0.0.1:10086, so AI agents can navigate, read, and interact with any website using the user's existing login sessions. ## Core Features & Use Cases - Full Browser Control: Navigate, click, fill forms (including ProseMirror/Lexical rich editors), take screenshots, save pages as PDF, and monitor network requests via simple HTTP commands. - Session-Based Tab Management: Group all tabs for a task under one named session and tab group, borrow the user's currently active tab, and close everything on request. - Accessibility-Tree Snapshots: Read page content and locate elements through semantic @e refs that survive CSS class changes, with raw CDP and JavaScript evaluation as escape hatches. - Use Case: Ask the agent to research a topic across multiple sites — it opens a labeled tab group, reads each page, fills search forms, captures screenshots, and compiles findings, all in your real browser. ## Quick Start Use the webbridge skill to open example.com in a new tab and take a screenshot of the page.
