What Knode is

Knode is your second brain for AI: a local memory that is yours, that every AI agent you use can read and add to, and that you see and organise in a native app.

• One memory for every agent. Claude Code, Codex, Gemini and Claude Desktop read and write the same memory through MCP, instead of each keeping its own.
• Your notes and your AI conversations, together. Ideas, decisions, notes and everything that comes out of each conversation, connected as nodes and links.
• Spaces with permissions. Work and personal apart. Each space is a .knode file, and you decide which agent sees which space and whether it can write.
• Local and open. In your files, on your Mac, in an open format. Every node is Markdown. No AI inside the app: the AI is the agent that connects to Knode.

Principles

These decide the hard calls. When two options look equal, pick the one that respects them best.

1. Yours. Your thoughts live in files you own, on your devices. No lock-in, no "our data".
2. Open. The .knode format is documented and stable; the code is open source.
3. Native. A core in Rust, and an app built with each platform's own tools, starting with macOS.
4. Local first. Everything works offline; anything that leaves your device is your choice, and encrypted end to end.
5. Calm. No noise, no streaks, no magic tricks. Knode helps you think; it doesn't perform.
6. Simple, then deep. Easy the first minute, powerful after a year.

Who it's for

People who work with AI agents every day and want what they and their agents learn to add up to something they can see, search and keep. The first user is the person building it.

Where it stands

Knode is in design and built in public. What comes next is in the roadmap; how it is built is in decision 0005 and decision 0006.