What problem does it solve? Manually scripting browser automation with fragile CSS selectors breaks whenever page layouts change, and raw DOM dumps overwhelm LLM context windows. This Skill provides a CLI that exposes stable element references (@e1, @e2) derived from accessibility trees, making browser control deterministic and readable for AI agents. ## Core Features & Use Cases - Element Ref Automation: Snapshot pages into accessibility trees with stable refs, then click, fill, type, and select elements by ref instead of brittle selectors. - Session Isolation: Run parallel browser workflows with named sessions, JSON output for scripting, and CDP connections to existing Chrome instances. - Network & Capture Tools: Mock API responses, block requests, record video, take screenshots, and export pages as PDF. - Use Case: Automate a login flow, navigate to a reports dashboard, extract table data as JSON, and capture a verification screenshot — all from shell commands an AI agent can reason about. ## Quick Start Ask the agent to open a URL with agent-browser, take an interactive snapshot to get element refs, then fill the login form and click submit.
