Your AI agents forget everything
between sessions.

Meshnote gives them a shared brain.

A knowledge base designed for LLM agents. They organize, synthesize, and maintain a wiki via MCP — you curate sources and browse.

How it works

Three steps to a compounding knowledge base

1

Drop sources in

Articles, papers, meeting notes, podcast transcripts — anything you want your agents to know. Plain files, no special format.

2

Your agent organizes

The LLM reads each source, creates entity and concept pages, cross-links with wikilinks, and maintains a structured wiki via MCP.

3

Every agent remembers

Claude Code, Cursor, Claude.ai — every agent reads the same brain. Knowledge compounds across sessions, tools, and devices.

Pricing

One product, two modes

Local
Free forever

Your machine, your files. No account needed.

  • CLI tool + local MCP server
  • Unlimited projects
  • Plain markdown, Obsidian-compatible
  • Works with any MCP client
Install CLI
Hosted
$8 /month

One brain, every agent, every device.

  • Cloud MCP endpoint
  • Cross-device, cross-agent
  • OAuth for Claude.ai web
  • API keys for agent connections
  • Same plain markdown — export anytime
Subscribe

Get started

Up and running in two minutes

Install the CLI, point your agent at it, start dropping sources.

1. Install

$ npm install -g meshnote

2. Create a brain

$ npx meshnote init ~/brains/research --template research

3. Tell your agent

{
  "mcpServers": {
    "meshnote": { "command": "npx", "args": ["meshnote", "mcp"] }
  }
}

Add to your Claude Code, Cursor, or any MCP client config.