Nanobrowser is an open-source Chrome extension that brings AI-powered web automation directly into your browser. Unlike cloud-based tools like OpenAI Operator, it runs entirely locally—your API keys and data never leave your machine. Designed for developers, power users, and privacy-conscious teams, Nanobrowser enables complex web tasks through a multi-agent system where specialized AI agents (Planner and Navigator) collaborate to navigate, extract, and act on web content. With support for multiple LLM providers including OpenAI, Anthropic, Gemini, Ollama, and custom OpenAI-compatible APIs, it offers flexibility in cost, performance, and privacy. Whether you’re automating research, data extraction, or dynamic form filling, Nanobrowser gives you full control over your automation stack without subscriptions or vendor lock-in.
What You Get
- Multi-agent System - Two specialized AI agents (Planner and Navigator) work together to break down complex web tasks into steps and execute them, with the Planner dynamically adjusting strategy when obstacles are encountered.
- Interactive Side Panel - A built-in chat interface provides real-time status updates, task progress, and the ability to ask follow-up questions about completed automation workflows.
- Multiple LLM Support - Connect and assign different models to Planner and Navigator agents using API keys from OpenAI, Anthropic (Claude), Gemini, Groq, Cerebras, Llama, or any OpenAI-compatible provider like Ollama.
- Local Model Support via Ollama - Run LLMs entirely on your machine using local models such as Qwen3-30B-A3B-Instruct-2507, Falcon3 10B, or Mistral Small 24B with zero API costs and maximum privacy.
- Task Automation - Automate repetitive web tasks like extracting headlines from news sites, finding trending GitHub repos, or comparing products on Amazon with simple natural language prompts.
- Conversation History - Access and manage past automation sessions to review outputs, reuse prompts, or refine workflows over time.
Common Use Cases
- Building a research assistant for competitive analysis - Automate gathering product features, pricing, and reviews from multiple e-commerce sites to compare competitors without manual copy-pasting.
- Automating data extraction for market intelligence - Extract top 10 trending Python repositories from GitHub with star counts and recent commits using a single natural language command.
- Problem → Solution flow: Need to monitor dynamic web content without APIs? - Nanobrowser navigates pages, parses DOM elements, and extracts structured data even from JavaScript-heavy sites like Amazon or TechCrunch, bypassing the need for official APIs.
- Team workflow: DevOps teams automating cross-platform monitoring - Use Nanobrowser to check status pages, login portals, or SaaS dashboards across browsers and extract alerts or uptime metrics into local logs without exposing credentials to third-party services.
Under The Hood
This project is an AI-powered web automation browser extension that enables intelligent, natural language-driven web interactions through a modular and extensible architecture. It combines browser automation with agent-based orchestration to deliver sophisticated web browsing capabilities within a Chrome extension framework.
Architecture
This codebase adopts a layered and modular architecture to support scalable development and clear separation of concerns.
- The system is organized into distinct modules for agents, browser interactions, UI components, and utilities, ensuring loose coupling and high cohesion.
- Core logic is structured using well-defined packages such as dev-utils, i18n, and HMR to handle cross-cutting concerns effectively.
- Design patterns like strategy and factory are applied in action and agent implementations, enabling flexible and extensible automation workflows.
- Component communication is managed through React-based UIs and background service workers, with a clear data flow architecture.
Tech Stack
Built primarily with TypeScript and modern JavaScript frameworks, the project leverages a robust tech stack for extension development.
- The codebase uses TypeScript for type safety and enhanced developer experience across the codebase.
- React is integrated for UI component development, while Turbo and pnpm manage a monorepo structure.
- The tech stack emphasizes automated builds, cross-platform compatibility, and integration with web extension APIs.
Code Quality
The codebase prioritizes security and robustness, particularly in handling untrusted content and preventing malicious patterns.
- A comprehensive testing approach covers sanitization logic and threat detection, ensuring reliable handling of external inputs.
- Error handling follows defensive programming practices with structured try/catch blocks and graceful degradation strategies.
- Code consistency is maintained through TypeScript usage and a well-structured module organization that enhances maintainability.
What Makes It Unique
This project introduces a unique synergy between AI-driven automation and browser-based interaction in a Chrome extension.
- It implements a modular agent architecture that orchestrates complex, multi-step browser tasks through structured planning and execution flows.
- The integration of natural language interaction with web automation sets it apart from conventional browser automation tools.
- Its extensible design allows for modular expansion of agent capabilities and interaction patterns without disrupting core functionality.