Open Source (RA-H OS)

Self-host your own RA-H contextual substrate with full source access

RA-H Open Source (RA-H OS)

A local contextual substrate — SQLite database with UI and MCP server. Your AI tools capture and connect your knowledge directly.

What This Does

  1. Stores knowledge locally — Notes, bookmarks, ideas, research in a SQLite database on your machine
  2. Provides a UI — Browse, search, and organize your nodes at localhost:3000
  3. Exposes an MCP server — Claude Code, Cursor, or any MCP client can query and add to your knowledge base

Your data stays on your machine. Nothing is sent anywhere unless you configure an API key.

Requirements

  • Node.js 20.18.1+nodejs.org
  • macOS — Works out of the box
  • Linux/Windows — Requires building sqlite-vec manually (see installation page)

Install

Use the ra-h_os README as the source of truth for installation and first-run setup. It is kept closer to the actual repo than this site and includes the current bootstrap steps.

At a high level:

  1. Clone ra-h_os
  2. Install dependencies
  3. Run the bootstrap script from the README
  4. Start the app
  5. Open chat or your MCP client and say let's get started

API Keys

Optional but recommended. Without a key, you can still create and organize nodes manually.

With OpenAI configured, you get:

  • Auto-generated descriptions when you add nodes
  • Automatic dimension/tag assignment
  • Semantic search (find similar content, not just keyword matches)

Cost: Less than $0.10/day for heavy use. Most users spend $1-2/month.

Setup: The app will prompt you on first launch, or go to Settings → API Keys.

Get a key at platform.openai.com/api-keys

Where Your Data Lives

~/Library/Application Support/RA-H/db/rah.sqlite   # macOS
~/.local/share/RA-H/db/rah.sqlite                  # Linux
%APPDATA%/RA-H/db/rah.sqlite                       # Windows

This is a standard SQLite file. You can:

  • Back it up by copying the file
  • Query it directly with sqlite3 or any SQLite tool
  • Move it between machines

Connect Claude Code (or other MCP clients)

Add to your ~/.claude.json:

{
  "mcpServers": {
    "ra-h": {
      "command": "npx",
      "args": ["ra-h-mcp-server"]
    }
  }
}

Restart Claude Code fully (Cmd+Q on Mac, not just closing the window).

Verify it worked: Ask Claude "Do you have rah_ tools available?" — you should see tools like rah_add_node, rah_search_nodes, etc.

What's Different from the Mac App?

RA-OS is a "lite" version focused on core knowledge management. Some features from the Mac app are not included:

FeatureMac AppRA-OS
Voice ModeYesNo
Built-in chat assistantYesNo
Auto-updatesYesManual (git pull)
Native desktop appYesWeb browser
Included AI creditsYesBYO API keys
Table ViewYesYes
Temporal QueriesYesYes (via MCP)

Key Features

FeatureDescription
Local SQLiteAll data stored locally in a single SQLite file
MCP ServerConnect Claude Code, Cursor, or Windsurf to your knowledge graph
Skills SystemShared markdown instructions for graph conventions and workflows
BYO API KeysUse your own OpenAI key, plus Anthropic if your setup needs it
MIT LicenseFork it, modify it, deploy it however you want

Next Steps