lg / notes

One web of ideas, shared by you and every agent you use

Notes is a Zettelkasten, the slip box with which the sociologist Niklas Luhmann wrote more than 70 books, kept as plain Markdown files in your own Google Drive, and open to your AI agents through MCP. Every agent starts from what is already known, adds what it concludes, and links it to the rest.

What it does

• One idea per note, linked with reasons. Each link says why two notes connect and how strongly, from 1 (loosely related) to 5 (one can't be understood without the other).
• Trains of thought. Every note follows the note it continues, so ideas keep their order (1, 1a, 1a1, 1b…) as in Luhmann's box.
• Deeper levels. When several notes add up to an insight, a deeper note states it and condenses them. You can look at your notebook one level at a time, or all at once.
• A live map. The connections, coloured from cold to hot by weight. The notebook computes which notes are the hubs, which ones bridge separate lines of thought, and how important each note is, using measures from network science.
• Papers, boards and files. Literature notes with their PDFs, whiteboards that lay notes out in space, pictures and recordings inside notes.
• Nothing is lost. Every save is a version you can compare and restore, and deleted notes wait in the trash.

Why it is leverage with agents

A chat with an AI forgets everything when it ends, and different agents never see each other's work. Notes gives all of them one memory that keeps growing.

• Memory beyond one conversation. An agent opens the notebook with one call (notes_overview). It sees what is known, what is open, what is missing and what is not connected yet, and it saves results, decisions and open questions as it goes, without being asked.
• Many agents, one notebook. Claude Code, Codex, Gemini, Grok, OpenCode, claude.ai or ChatGPT can all read and write the same notes. Each note says which agent wrote it, so you can let several agents work on one question, compare what they conclude, and keep the result.
• Structure, not a pile. Agents follow the whole method: one idea per note, placed in its train of thought, linked with weighted reasons, condensed into deeper notes, and drawn as boards when a picture explains better. The notebook shows you where your ideas concentrate and where separate lines of thought meet.
• Safe to let agents work on their own. Every change is signed and kept in History, so agents can tidy, merge and rewrite, and you can undo anything. Each agent has its own key, read only or able to write, that you can revoke at any time.
• Standard and current. The MCP server lives at notes.lovegrover.com/mcp, with OAuth sign-in. Connecting an agent takes one click, or one setup app for every agent on a computer. The standing instruction agents follow carries a version number, and agents update their own copy when the method changes.

How to start

1. Open notes.lovegrover.com and continue with Google. Your notes go into a folder called Notes in your Drive.
2. Under Agents, connect each agent with one click, or download the Setup app to connect every agent on your computer at once.
3. Work as usual. Ask an agent to read your notes before it starts, or let it save what you conclude together.

Your data

Your notes are files in your own Google Drive, and the app can reach only the files it created there. This site keeps no copy of them, does not sell or advertise with anything, and does not use your notes to train AI models. The details are in the privacy policy.