Get Started
Get Started
RushDB is a graph database for apps and AI agents: push JSON, get connected records you can filter, traverse, aggregate and search by keywords or meaning. It runs in one of two ways, and the SDKs, SearchQuery and MCP server work the same with both.
| RushDB Cloud | Self-hosted engine | |
|---|---|---|
| What it is | The managed service: the RushDB engine plus the dashboard, managed embeddings and AI features | The Apache-2.0 rushdb binary, on your machines |
| Start | Sign up at app.rushdb.com and create an API key | Install and run rushdb up |
| Connect with | The API key | The engine URL and token (SDK direct mode) |
| Choose it when | You want nothing to operate, managed embeddings, natural-language search, or a hosted MCP connector | Data must stay on your infrastructure, or you want SQL, Bolt, gRPC and the engine's other interfaces |
Not sure? Start on Cloud; the same code runs against your own engine later. See RushDB Cloud or a self-hosted engine for the full comparison.
1. Get a database
- RushDB Cloud
- Self-hosted engine
- Sign up at app.rushdb.com and create a project.
- Create an API key in the project, and export it:
export RUSHDB_API_KEY="your-api-key"
Install the engine and start a node with a persistent data directory (Homebrew, Docker, Cargo and release archives are in Installation):
curl -fsSL https://raw.githubusercontent.com/rush-db/rushdb/main/install.sh | sh
export RUSHDB_TOKEN="$(openssl rand -hex 32)"
rushdb up --data-dir ./rushdb-data
The node listens on http://127.0.0.1:3000. Check it:
curl http://127.0.0.1:3000/health
2. Connect
| Path | Best for | Guide |
|---|---|---|
| SDK | Applications, services and data pipelines | Quick tutorial |
| Local MCP server | Claude Desktop, Cursor, VS Code | MCP |
| Hosted MCP connector [Cloud] | ChatGPT, Claude.ai | MCP |
| Agent skills | Teaching an agent to use RushDB well | Agent Skills |
| Native agent memory | OpenClaw and Hermes | Agents |
| SQL, Cypher, gRPC, native protocol | A self-hosted engine | SQL and Cypher clients |
Concepts to know
- Records — a node with a label and any JSON properties. No schema required upfront.
- Labels — the record's type (
SESSION,DECISION,ARTICLE). You define them by pushing data. - Relationships — named, directed edges between records. Created from nested JSON, or explicitly.
- SearchQuery — one JSON shape to filter, traverse, sort, page and aggregate records.