BrowserSmith automates four ChatGPT tabs to plan, write, review and run real apps
SouthernArcher9793 · reddit · 2026-07-27
BrowserSmith 1.0 automates four ChatGPT tabs into a small build team: one plans, one writes, one reviews with a PRINT/RETRY gate, and one audits the finished project.
It goes beyond generation and actually runs the app: it installs dependencies, starts the dev server or entry point, screenshots the result, and patches errors from the framework overlay instead of regenerating files blindly.
- No API key needed: it uses the ChatGPT web UI, so a free account works.
- Supports Next.js, Vite, static sites, Node CLIs, Python, and an Auto mode that picks the stack.
- Verified on a pomodoro app and a CSV stats CLI; 109 tests pass on Windows and Ubuntu.
- Limits: ChatGPT UI changes can break it, free-tier rate limits are slow, and npm install executes the model-generated package.json in the project folder.
- Privacy claims: no telemetry, temporary chats, local session storage, cookie names/expiry only.
More from coding & agent
- Cursor should add a direct Grok Build backend, says a new ACP integration proposal — Daniel_Farinax · 2026-07-27
- Berkeley’s ABBEL trains agents to store graded belief states instead of full history — berkeley_ai · 2026-07-27
- AI Stupid Level says it tracks silent model drift across 20+ providers and 98,000 MAU — ionutvi · 2026-07-27
- Browser-agent builder asks whether website drift still breaks self-healing workflows — ryicean · 2026-07-27
- Testing Claude to Automate B2B Lead Gen and Booking — petewoodbridge · 2026-07-27
- Audited 50+ AI agent skills, and most had dangerous command-execution risks — aiz27 · 2026-07-27