talk2browser

Learn, contribute, and automate browser tasks with natural language.
An open source project for the future of browser automation.

Control and test web apps using plain English.

Generate Playwright, Cypress, and Selenium scripts.

Easily extend with new browser or script tools.

YOLOv11-based UI element detection and metadata extraction.

Manage secrets securely via environment variables.

Flexible orchestration for complex browser tasks.

Add new tools, LLMs, or workflows as Python modules.

Generate Comprehensive PDF Documentation and Reports.

The Architecture

●You start with a simple command to our CLI interface – just plain English describing your task.
●Our AI agent processes this, leverages the LLM engine to understand and plan the required actions.
●Then it executes those actions in a real browser – not simulation, actual browser interaction. Every action is recorded by our action recorder.
●Finally, our script generator takes that action sequence and creates clean, readable code. Whether you need Selenium for legacy systems, Playwright for modern apps, or Cypress for frontend testing.
●Notice the vision detection component – that's what lets our AI "see" the page and interact with elements intelligently, even when the DOM is complex.

Once your automation script is ready, it can be repeated as many times as needed with consistent, reliable results every single time.

✔️Instant execution without any setup delays or configuration changes

✔️Unlimited repeats to run the same automation hundreds or thousands of times

✔️Consistent results with the same reliable outcome every execution

✔️Schedule & automate to set up regular runs for recurring workflows

Join the open source movement and automate your browser with natural language.