rushdb

---

Quick start

curl -fsSL https://raw.githubusercontent.com/rush-db/rushdb/main/install.sh | sh
rushdb up

The script installs the rushdb binary into ~/.rushdb/bin after checking the release checksum (and its Sigstore signature when cosign is installed). rushdb up starts a durable node on http://127.0.0.1:3000 with its data in ./.rushdb and the local token local-token.

Write a nested JSON document. Every nested object becomes its own record, linked to its parent:

export RUSHDB_TOKEN=local-token
curl -s http://127.0.0.1:3000/protocol/v1/execute \
  -H "Authorization: Bearer $RUSHDB_TOKEN" -H 'Content-Type: application/json' \
  -d '{"protocolVersion":1,"operation":{"kind":"ingest","ingest":{"kind":"documents","label":"COMPANY",
       "documents":{"name":"Acme","DEPARTMENT":[
         {"name":"Engineering","headcount":12,"EMPLOYEE":[{"name":"Ada","role":"Staff Engineer"},{"name":"Linus","role":"Engineer"}]},
         {"name":"Sales","headcount":4,"EMPLOYEE":[{"name":"Grace","role":"Account Executive"}]}]}}}}'

Then find the engineers at Acme. The filter reaches across relationships by naming the related labels:

curl -s http://127.0.0.1:3000/protocol/v1/execute \
  -H "Authorization: Bearer $RUSHDB_TOKEN" -H 'Content-Type: application/json' \
  -d '{"protocolVersion":1,"operation":{"kind":"query","query":{"kind":"records","query":{
       "labels":["EMPLOYEE"],
       "predicate":{"role":{"$contains":"Engineer"},"DEPARTMENT":{"COMPANY":{"name":"Acme"}}}}}}}'

Continue with the Quickstart.

Query it your way

The same records answer every interface.

SQL over the PostgreSQL wire protocol, with psql or any PostgreSQL driver. Labels are tables:

PGPASSWORD=$RUSHDB_TOKEN psql "host=127.0.0.1 port=5432 user=rushdb dbname=rushdb sslmode=disable" \
  -c "SELECT name, headcount FROM department ORDER BY headcount DESC"

Cypher over HTTP, or over Bolt with the Neo4j drivers (rushdb up --languages sql,cypher --bolt-port 7687):

MATCH (d:DEPARTMENT)-->(e:EMPLOYEE)
RETURN d.name AS department, count(e) AS people ORDER BY people DESC

The SDKs in direct mode (npm install @rushdb/javascript-sdk):

import RushDB from '@rushdb/javascript-sdk'

const db = new RushDB(process.env.RUSHDB_TOKEN, { direct: true, engineUrl: 'http://127.0.0.1:3000' })

const engineers = await db.records.find({
  labels: ['EMPLOYEE'],
  where: { role: { $contains: 'Engineer' }, DEPARTMENT: { COMPANY: { name: 'Acme' } } },
  orderBy: { name: 'asc' }
})
console.log(engineers.data.map((employee) => employee.data.name)) // [ 'Ada', 'Linus' ]

Gremlin, SQL/PGQ GRAPH_TABLE, Mongo-style documents and ISO GQL use the same HTTP endpoint. See Query languages.

What’s in the box

Capability	What it means
No schema to design	Write any JSON shape. Types are inferred per value, nested objects become linked records, and the engine keeps an exact catalog of labels, properties and relationships that you can query.
Properties are first-class	The Labeled Meta-Property Graph (LMPG) stores each property as a typed definition shared by the records that use it, so schema discovery, value ranges and facets come from the storage layout itself.
Many query languages	A native JSON query language, SQL, Gremlin, SQL/PGQ, Mongo-style documents, and openCypher and ISO GQL as release candidates.
Search	Vector indexes (exact and HNSW) over the vectors you supply, BM25 full-text indexes, and hybrid search that fuses both.
Durable and transactional	A write-ahead log, MVCC snapshots, multi-statement transactions, backups with point-in-time restore, and streaming replication to read-only followers.
Standard protocols	HTTP, the PostgreSQL wire protocol, gRPC and Bolt from one process.
Multitenancy	Optional fail-closed tenants, each with its own keyspace.
One binary	rushdb serve runs the server; rushdb up runs a local node. No other service to operate.

Deploy

Where	How
Docker	docker run -p 127.0.0.1:3000:3000 -v rushdb:/data -e RUSHDB_TOKEN=change-me rushdb/rushdb
Docker Compose	deploy/templates/docker-compose, with an HTTPS variant behind Caddy
Kubernetes	Helm chart: helm install rushdb oci://ghcr.io/rush-db/charts/rushdb --set auth.token=...
Linux VM	cloud-init for DigitalOcean, Hetzner, AWS, GCP, Azure; systemd unit
Platforms	Fly.io, Render, Coolify, Dokploy

Other ways to install

Method	Command	Notes
Install script	curl -fsSL https://raw.githubusercontent.com/rush-db/rushdb/main/install.sh | sh	Linux (glibc 2.28+) and macOS, x86-64 and arm64. sh -s -- --version v3.0.0 pins a release.
Homebrew	brew install rush-db/tap/rushdb	macOS and Linux. brew services start rushdb runs it in the background.
Docker	rushdb/rushdb	linux/amd64 and linux/arm64; also ghcr.io/rush-db/rushdb.
Cargo	cargo binstall rushdb-server or cargo install rushdb-server --locked	binstall fetches the release binary; install compiles (needs a C++ compiler).
Release archive	GitHub releases	Archives with SHA-256 checksums, an SBOM, and Sigstore signatures.
Source	cargo build --release -p rushdb-server --bin rushdb	See Installation.

Windows is not supported natively; use WSL 2 or Docker. Read Security before exposing a node beyond loopback.

Status

Area	Status
Native protocol, SQL, Gremlin text traversals, Mongo-style documents	Stable
SQL/PGQ GRAPH_TABLE	Stable, read-only
openCypher, ISO GQL, Bolt	Release candidate, opt-in: the external conformance suites do not pass in full (details)
PostgreSQL wire	SQL reads and single-statement writes; checked with psql, node-postgres, psycopg and tokio-postgres, not every client or ORM
Replication	Asynchronous streaming to read-only followers; promotion is manual, with no automatic failover
Raft, storage sharding	Preview, not for production: Raft runs in one process with in-memory state; multi-shard commits are not crash-atomic

The manual documents each surface’s exact limits.

Connect

Surface	Default	Enable
HTTP native protocol	127.0.0.1:3000, POST /protocol/v1/execute	always on
PostgreSQL wire	127.0.0.1:5432	on with sql (default); --pgwire-port 0 turns it off
gRPC (rushdb.v1.Database/Execute)	off	--grpc-port
Bolt 5.0	off	--bolt-port with cypher

The JavaScript and Python SDKs, the MCP server, the agent skills and the LangChain and LangGraph integrations (langchain-rushdb, @rushdb/langchain) work with a self-hosted engine in direct mode.

Self-hosted engine and RushDB Cloud

This engine is the part of RushDB you self-host. It does not generate embeddings, host a dashboard, or manage users: those come with RushDB Cloud, the managed service that runs this engine and adds the dashboard, managed embeddings and the AI features (natural-language search, relationship-pattern suggestions). The SDKs and the MCP server work with either: give them a RushDB Cloud API key, or point them at your engine (direct: true, engineUrl) for everything except the Cloud features, which they refuse with CloudOnlyError.

Documentation

• Manual: getting started, data model, queries, languages, search, operations, and reference tables.
• docs.rushdb.com: tutorials, the SDKs and the MCP server.
• Upgrading from RushDB 2.x: 2.x was the Neo4j-based stack, now on the 2.x branch of this repository.

Versions

• RushDB 3.x (main): the current version.
• RushDB 2.x (legacy): the Neo4j-based stack, on the 2.x branch and the rushdb/platform image. Security fixes only until 2027-10-28.

Contributing

See CONTRIBUTING.md. The quality gate is:

cargo test --workspace --no-fail-fast --locked
cargo clippy --workspace --all-targets --locked
cargo fmt --all -- --check
scripts/check-quality-bar.sh

• Security policy
• Release process
• Code of conduct
• Changelog

Licensed under Apache-2.0.