# NornicDB — Graph+Vector DB for Agent Memory

> NornicDB is a Neo4j-compatible graph+vector database for agent memory and GraphRAG; run it in Docker and manage it from a localhost admin UI.

## Install

Save the content below to `.claude/skills/` or append to your `CLAUDE.md`:

## Quick Use

```bash
docker run -d --name nornicdb -p 7474:7474 -p 7687:7687 -v nornicdb-data:/data timothyswt/nornicdb-amd64-cpu-bge:latest
# then open:
open http://localhost:7474
# connect via Bolt:
# bolt://localhost:7687
```

## Intro

NornicDB is a Neo4j-compatible graph+vector database for agent memory and GraphRAG; run it in Docker and manage it from a localhost admin UI.

**Best for:** GraphRAG builders who want graph traversal + vector retrieval in one engine

**Works with:** Docker, Neo4j drivers (Bolt/Cypher), REST/GraphQL/gRPC (per README)

**Setup time:** 5-15 minutes

### Key facts (verified)

- GitHub: 727 stars · 38 forks · pushed 2026-05-13.
- License: MIT · owner avatar + repo URL verified via GitHub API.
- README-verified entrypoint: `docker run -d --name nornicdb -p 7474:7474 -p 7687:7687 -v nornicdb-data:/data timothyswt/nornicdb-arm64-metal-bge:latest`.

## Main

- Use Docker for the quickest evaluation; the README lists separate images for Apple Silicon, CPU-only amd64, CUDA, and Vulkan environments.

- Treat Neo4j compatibility as the migration path: start by connecting your existing Neo4j driver to `bolt://localhost:7687` and keep queries in Cypher.

- Use the admin UI at `http://localhost:7474` to validate schema/indexes and iterate before wiring it into a production GraphRAG pipeline.

### Source-backed notes

- README Quick Start shows Docker commands mapping ports 7474 (admin UI) and 7687 (Bolt) and points to `http://localhost:7474`.
- README states it is Neo4j-compatible (Bolt + Cypher) and also exposes REST/GraphQL/gRPC interfaces.
- README notes different images/profiles (arm64/amd64, CPU/CUDA/Vulkan) and provides a from-source build path using Go.

### FAQ

- **Is it compatible with Neo4j clients?**: README says Bolt + Cypher compatibility is a core goal for drop-in usage.
- **Which port is the admin UI?**: 7474 — README links to `http://localhost:7474`.
- **Do I need a GPU?**: No — README includes CPU-only images; GPU paths are optional for acceleration.

## Source & Thanks

> Source: https://github.com/orneryd/NornicDB
> License: MIT
> GitHub stars: 727 · forks: 38

---

<!-- ZH -->

## Quick Use

```bash
docker run -d --name nornicdb -p 7474:7474 -p 7687:7687 -v nornicdb-data:/data timothyswt/nornicdb-amd64-cpu-bge:latest
# then open:
open http://localhost:7474
# connect via Bolt:
# bolt://localhost:7687
```

## Intro

NornicDB 是面向 Agent Memory/GraphRAG 的图+向量数据库，兼容 Neo4j Bolt/Cypher；可用 Docker 启动，并在 localhost 管理 UI 上操作（端口 7474/7687）。

**Best for:** 需要“图遍历 + 向量检索”一体化的 GraphRAG/知识系统团队

**Works with:** Docker；Neo4j driver（Bolt/Cypher）；REST/GraphQL/gRPC（README）

**Setup time:** 5-15 minutes

### Key facts (verified)

- GitHub：727 stars · 38 forks；最近更新 2026-05-13。
- 许可证：MIT；作者头像与仓库链接均已通过 GitHub API 复核。
- README 中核对过的入口命令：`docker run -d --name nornicdb -p 7474:7474 -p 7687:7687 -v nornicdb-data:/data timothyswt/nornicdb-arm64-metal-bge:latest`。

## Main

- 评估阶段优先用 Docker；README 列出 Apple Silicon/amd64 CPU/CUDA/Vulkan 等多种镜像，方便按硬件选型。

- 把 Neo4j 兼容当迁移通道：先用现有 Neo4j driver 连 `bolt://localhost:7687`，继续用 Cypher 跑起来再逐步替换。

- 用 `http://localhost:7474` 的管理 UI 验证 schema/index/向量能力，跑通后再接入你的 GraphRAG 流水线。

### Source-backed notes

- README 的 Quick Start 给出 Docker 启动命令，映射 7474（管理 UI）与 7687（Bolt）端口，并提示访问 `http://localhost:7474`。
- README 说明它兼容 Neo4j（Bolt + Cypher），同时提供 REST/GraphQL/gRPC 接口。
- README 描述多镜像配置（arm64/amd64、CPU/CUDA/Vulkan），并提供用 Go 从源码构建的路径。

### FAQ

- **能用 Neo4j 客户端/驱动吗？**：README 写明支持 Bolt + Cypher，目标就是尽量做到 drop-in 兼容。
- **管理 UI 在哪个端口？**：7474。README 提示访问 `http://localhost:7474`。
- **必须要 GPU 吗？**：不需要。README 提供 CPU-only 镜像；GPU 只是可选加速路径。

## Source & Thanks

> Source: https://github.com/orneryd/NornicDB
> License: MIT
> GitHub stars: 727 · forks: 38


---
Source: https://tokrepo.com/en/workflows/nornicdb-graph-vector-db-for-agent-memory
Author: AI Open Source