Back to Ecosystem Pulse
ToolProduction Ready

nanobrowser

by nanobrowser

Run multi-agent web automation workflows directly in Chrome

TypeScript
Updated Nov 24, 2025
Share:
12.2k
Stars
1.3k
Forks

View on GitHub

How It Works

Automates web tasks by running multi-agent workflows inside a Chrome extension using your own LLM API key. Chains and coordinates agents in the browser to perform scripted navigation, data extraction, and interactions, aligning with the Model Context Protocol (MCP) Pattern and the A2A Protocol Pattern. Notable for running automation locally in-browser (no backend required) and easy integration with Playwright and n8n flows.

Key Benefits

As agents take on more real-world web tasks, observing their behavior where actions happen—the browser—becomes essential for trust and debugging. Nanobrowser surfaces agent decisions, steps, and web actions in their native environment, making it easier to reproduce failures and audit delegation. This supports Human-in-the-Loop practices for pre-production testing and runbooks.

When to Use

Developers and builders who need to prototype, test, and run multi-agent browser automations without a backend. This approach aligns with the Planning Pattern to structure workflows and improve reliability.

How It's Used

  • Automate complex web workflows by delegating subtasks to specialized agents in the browser
  • Reproduce and debug agent failure modes with browser-level action logs and step traces
  • Prototype agent-to-agent delegation patterns and integrate results into n8n or Playwright pipelines
Works With
openaiplaywrightn8nchrome-extension
Topics
agentaiai-agentsai-toolsautomationbrowserbrowser-automationbrowser-usechrome-extensioncomet+10 more
Similar Tools
openai-operatorn8n
Keywords
multi-agent orchestrationweb-automationagent delegationbrowser-extension