RA-H Light

A lightweight local knowledge graph UI with MCP server. Connect your AI coding agents to a personal knowledge base. BYO API keys, no cloud dependencies.

What is RA-H Light?

RA-H Light is a stripped-down version of RA-H focused on being a knowledge management backend for AI agents. It provides:

• 2-panel UI – Nodes list + focus panel for viewing/editing knowledge
• SQLite + sqlite-vec – Local vector database with semantic search
• MCP Server – Connect Claude Code, Cursor, or any MCP-compatible AI assistant
• Workflows – Editable JSON workflows for multi-step operations

What's removed: Built-in chat agents, voice features, delegation system. RA-H Light is designed for technical users who want to bring their own AI agents via MCP.

Platform Support

Platform	Status
macOS (Apple Silicon)	Supported
macOS (Intel)	Supported
Linux	Requires manual sqlite-vec build
Windows	Requires manual sqlite-vec build

Quick Start

git clone https://github.com/bradwmorris/ra-h_os.git
cd ra-h_os
npm install
npm rebuild better-sqlite3
scripts/dev/bootstrap-local.sh
npm run dev

Open http://localhost:3000 → Settings → API Keys → add your OpenAI key (for embeddings).

Connecting AI Agents via MCP

RA-H Light exposes an MCP server that external AI assistants can use to read/write your knowledge graph.

Claude Code Integration

Add to your Claude Code settings:

{
  "mcpServers": {
    "rah": {
      "command": "node",
      "args": ["/path/to/ra-h_os/apps/mcp-server/stdio-server.js"]
    }
  }
}

Available MCP Tools

Tool	Description
rah_add_node	Create a new knowledge node
rah_search_nodes	Search nodes by text
rah_update_node	Update an existing node
rah_get_nodes	Get nodes by ID
rah_create_edge	Connect two nodes
rah_query_edges	Find connections
rah_update_edge	Update a connection
rah_create_dimension	Create a tag/category
rah_update_dimension	Update a dimension
rah_delete_dimension	Delete a dimension
rah_search_embeddings	Semantic vector search

HTTP MCP Server

For non-stdio clients, start the HTTP server:

node apps/mcp-server/server.js

Listens on http://127.0.0.1:44145/mcp by default.

Project Layout

app/                 Next.js App Router
src/
  components/        UI components
  services/          Database, embeddings, workflows
  tools/             Available tools for workflows
apps/mcp-server/     MCP server (stdio + HTTP)
docs/                Local documentation
scripts/             Dev helpers
vendor/              Pre-built binaries (sqlite-vec)

Commands

Command	Description
npm run dev	Dev server at localhost:3000
npm run build	Production build
npm run type-check	TypeScript validation
npm run sqlite:backup	Database snapshot
npm run sqlite:restore	Restore from backup

Documentation

• docs/0_overview.md – System overview
• docs/2_schema.md – Database schema
• docs/8_mcp.md – MCP server details

Linux/Windows Setup

The bundled sqlite-vec binary is macOS-only. For other platforms:

1. Clone https://github.com/asg017/sqlite-vec
2. Build for your platform
3. Place at vendor/sqlite-extensions/vec0.so (Linux) or vec0.dll (Windows)
4. Set SQLITE_VEC_EXTENSION_PATH in .env.local

Without sqlite-vec: UI, node CRUD, and basic search still work. Vector/semantic search requires it.

Contributing

See CONTRIBUTING.md. Issues and PRs welcome.

License

MIT