Introduction

NevoFlux is an AI-native browser. A built-in AI agent lives in the sidebar and can read the page you are on, drive the browser for you, control your computer, run skills, build interactive Canvas apps, and remember what matters in a local knowledge base. Don't just browse. Command.

How it fits together

NevoFlux is two pieces that ship together:

• The browser — a Firefox-based browser with an AI sidebar, a nevoflux:// settings/pages system, and the Canvas runtime for rich, interactive output.
• The agent — a local background process (the "daemon") bundled inside the browser. It runs the LLM conversation, calls tools, executes skills, and manages your knowledge base. It is installed and started automatically by the browser — you do not run it yourself.

The two talk over a secure local bridge. Your API keys, sessions, and knowledge base stay on your machine.

What the agent can do

• Browse for you — navigate, click, type, scroll, screenshot, and read pages.
• Control your computer — screenshots, mouse, and keyboard (platform dependent).
• Run skills — reusable capabilities like /brain, /app, and /video.
• Build Canvas apps — generate interactive HTML/React/Vue/Svelte apps, charts, diagrams, slides, and even video, rendered live in a tab.
• Remember — capture and recall knowledge in a local GBrain knowledge base.
• Use your tools — connect external MCP servers and define a tool whitelist.

This documentation describes features that exist in the shipping code. Where a capability is platform-specific or optional, the relevant page says so.