Nanobrowser

Category: Consumer / AI Browser Automation

Pricing: Free (bring your own LLM API key; no subscription fees)

Official site: https://nanobrowser.ai

GitHub: https://github.com/nanobrowser/nanobrowser (13k+ stars)

---

What it is

Nanobrowser is an open-source Chrome extension that runs a multi-agent AI workflow directly inside your browser. It is a self-hosted alternative to OpenAI Operator. You supply your own LLM API keys (OpenAI, Anthropic, Gemini, Ollama, Groq, Cerebras, or any compatible provider) and Nanobrowser runs three specialized agents locally:

• Planner - breaks down the user's natural-language task into steps
• Navigator - drives the browser (clicks, fills forms, scrolls)
• Validator - checks whether each step succeeded before proceeding

Because all execution happens inside your own Chrome session, your credentials and session cookies never leave your machine. There is no cloud proxy and no per-seat subscription.

Key facts

Property	Value
Type	Chrome extension (no CLI, no npm package)
Headless support	No - runs in your visible Chrome session
Proxy support	No built-in proxy layer
CAPTCHA solving	No
Anti-detect / stealth	N/A (your real Chrome profile)
LLM providers	OpenAI, Anthropic, Gemini, Ollama, Groq, Cerebras
Platform	Chrome / Chromium only
Open source	Yes (MIT)

When to use Nanobrowser

Use Nanobrowser when you need lightweight browser task automation driven by natural language and you are comfortable supplying your own LLM API key. It is best for one-off research tasks, data extraction from public pages, and exploratory workflows where you want to watch the agent act in your real browser.

Do NOT use it for headless/server-side automation, high-volume crawling, or workflows requiring stealth/proxy routing. For those jobs route to Steel, Browserbase, or Hyperbrowser instead.