Here is the link to VS Code extension - https://marketplace.visualstudio.com/items?itemName=scrapeco...
I've built browser automations for more than 5 years, and the constant frustration was always the sheer friction involved in getting working code – especially when debugging in headless mode or connecting to remote browsers.
When I started using LLMs to generate automation code, I found myself stuck in a repetitive loop: navigate to the desired page state, copy-paste HTML into the AI chat, and ask it to generate code. The worst part is that there was no easy way to run that generated code without losing the page state, forcing me to restart the browser session constantly. This wasted large amounts of time and mental energy. I built ScrapeCopilot to make this workflow seamless.
How it works:
ScrapeCopilot combines the power of a Jupyter-style notebook with a live Puppeteer browser session and integrated AI.
- Live Interactive Development: When you create an automation notebook, it initiates a fresh Puppeteer browser session. The page object is exposed directly to your notebook cells, allowing you to run any Puppeteer code against the live browser state and see the results instantly.
- AI-Powered Assistance: It integrates with GitHub Copilot (via the @scrapecopilot chat participant). The AI automatically sees the current page HTML, allowing it to generate highly relevant Puppeteer code based on your instructions directly within the chat.
- LLM Code Export: Once you've developed your automation logic, you can easily export the final, complete Puppeteer script based on your instructions.
This tool saves me hours daily, but even more importantly, it improves the developer experience in browser automation which is frustrating area.
I believe ScrapeCopilot can complement existing browser automation tools and frameworks by providing an interactive AI-assisted development experience.
Current Status & Future Plans:
- The extension currently works within VS Code. It will work in Cursor, but without chat support initially. I'm actively working on integrating a backend server to enable full chat functionality with Cursor.
- Currently the key workflow assumes that you create a new browser automation step by step, using code cells. But in my work I spend half of the time fixing existing automations, so my focus now is trying to adapt extension for debugging and fixing existing code.
- Playwright support is also on the list.
Check out short videos: - Demo: Headless False - https://scrapecopilot.ai/assets/demo-headless-false-Dhc_jeNR... - Demo: Headless True - https://scrapecopilot.ai/assets/demo-headless-true-PRQndDxP....
I'd love to hear your thoughts, feedback, and any suggestions!