Skip to main content

Introduction

Browsr is a headless browser agent: design scraping sequences with AI, then replay the same commands deterministically. Sessions stay warm, commands are explicit JSON payloads, and extraction endpoints return Markdown/HTML for downstream models.

What you get​

  • Design scraping sequence using AI
  • Execute sequences at scale
  • Export structured or LLM-friendly data

Installation​

Reach out for early access. Currently in beta.

Running​

browsr

Powered by Distri.

Local debugging​

Use vLLora to debug and fine-tune the system prompt, iterate locally, and try additional tools while you design sequences and selectors.

Next steps​

  • Configure Browsr via browsr.toml (see Configuration).
  • Customize the agent behavior by editing agents/browsr.md (see Configuration and the agents reference).
  • Review licensing (see License).
  • Keep sessions alive, scrape slices of the DOM, or ship the saved sequence—see the rest of these docs for focused guides.