What problem does it solve?
Automates repetitive, error-prone browser tasks inside the user's existing Chrome session so the agent can reuse logged-in state, avoid re-authentication, and perform real interactions without launching separate browser profiles.
Core Features & Use Cases
- Reuse existing Chrome: Connects to the user's real Chrome process via agent-browser and --auto-connect so cookies and sessions are preserved.
- Deterministic interactions: List tabs, snapshot interactive elements, click, fill, type, press keys, take screenshots, evaluate JS, and verify results after each step.
- Recording replay & cross-platform workflows: Parse and replay DevTools/Puppeteer/@puppeteer/replay recordings and run sequential subagents for publishing across social and marketing platforms.
- Robust edge-case handling: Iframe workarounds, contenteditable vs textarea handling, fallback selection strategies, and explicit user confirmations before destructive actions.
Quick Start
Use the chrome-automation skill to connect to your running Chrome with agent-browser --auto-connect, list tabs to reuse a logged-in page, and perform the requested navigation or interaction.