What problem does it solve? Showing a website or rendered HTML to a user normally requires leaving the terminal, and automating an already-open browser tab requires a separate driver setup. This Skill runs a real browser directly inside a terminal pane and exposes snapshot, click, fill, and eval commands against any open tab. ## Core Features & Use Cases - In-Terminal Browsing: Open URLs, localhost ports, or local HTML files in the current pane or a new split pane (right, left, down, up) with configurable size. - Tab Discovery: List running browsers and their tabs with terminal-browser ls, including JSON output with CDP ports and pane ids. - Agent-Compatible Automation: Drive any open tab through the agent-browser compatible terminal-browser action subcommand for snapshot, click, fill, and eval operations. - Use Case: Build an HTML report, open it with terminal-browser open ./report.html --split right to show it beside the conversation, then use terminal-browser action -- snapshot to verify the rendered content. ## Quick Start Open a browser pane beside this conversation showing localhost:3000, then take a snapshot of the page to inspect its elements.
