HelixDB/helix-db

HelixDB – A Graph Database with Built‑in Vector & Full‑Text Search

What it is – HelixDB is an open‑source (Apache 2.0) graph database written in Rust. It stores entities, relationships, embeddings, and text together in a single transactional engine and lets you query all three kinds of data in one request. It ships a server (run locally or in the managed Helix Cloud) and language SDKs for Rust, TypeScript, Python, and Go.

Why it matters for AI – Modern AI applications often need to combine a knowledge graph (entities & relations) with vector similarity (embeddings) and keyword search. HelixDB provides:

• Labeled property graph model for classic graph queries.
• Approximate nearest‑neighbor (ANN) vector search that can be filtered by graph traversals, enabling “find the most similar items among a subset of nodes”.
• BM25 full‑text search over node and edge properties.
• ACID‑guaranteed transactions on top of object‑storage‑backed durability.

Core features

Feature	Detail
Graph	Labeled property graph with CRUD via a Cypher‑compatible API (POST /v2/cypher).
Vector	ANN search (HNSW‑style) that can be combined with graph filters.
Full‑text	BM25 keyword search on any property.
Storage	Built on object storage; single‑writer commit model; optional in‑memory mode for fast dev.
Embedded mode	Run the database as a library inside your own process – no separate server needed.
Helix Cloud	Managed, high‑availability deployment with auto‑scaling read nodes.
CLI	helix command‑line tool for project scaffolding, local dev, cloud auth, and even an AI‑assistant “chef” that can generate a starter app.
SDKs	Uniform JSON query format; SDKs for Rust (helix-db crate), TypeScript (@helix-db/helix-db), Python (helix-db pip package), and Go (github.com/helixdb/helix-db/sdks/go).

Getting started

1. Install the CLI (macOS/Linux): curl -sSL https://install.helix-db.com | bash (Windows PowerShell uses the provided script).
2. Create a project: helix init local – writes a helix.toml config and example query.
3. Run locally: helix start dev (adds --disk to persist data).
4. Query: helix query dev --file examples/request.json or call the HTTP endpoint POST http://localhost:6969/v2/query from any SDK.
5. Optional AI‑assisted scaffolding: helix chef – the CLI invokes a coding agent (Claude, OpenAI Codex, etc.) to build a full app from a natural‑language description.

Example query (Rust)

let client = Client::new(None)?; // defaults to http://localhost:6969
let resp = client.query(add_user("John Doe".to_string())?).send().await?;
println!("{:?}", resp);

Similar snippets are provided for TypeScript, Python, and Go in the README.

Ecosystem & community

• Docs: https://docs.helix-db.com – quick‑start, SDK setup guides, Cypher reference, code‑base map.
• Discord: active chat for support and feature discussion.
• Helix Cloud: managed service for production workloads; CLI commands helix auth login, helix init cloud, helix query production ….
• License: Apache 2.0 (permissive, commercial‑friendly).

Who should use it – Teams building AI‑augmented applications that need a unified store for graph data, embeddings, and searchable text (e.g., recommendation engines, RAG pipelines, knowledge‑graph‑backed chatbots) and want a Rust‑first, open‑source solution with optional managed hosting.