# Neo4j MCP Servers — Cypher + Memory + Aura API > neo4j-contrib/mcp-neo4j bundles Neo4j MCP servers (Cypher, memory, Aura API); verified 946★ and documents stdio/HTTP/SSE plus timeouts and token limits. ## Install Merge the JSON below into your `.mcp.json`: ## Quick Use ```bash # Build server image locally (Cypher server subdir): git clone https://github.com/neo4j-contrib/mcp-neo4j.git && cd mcp-neo4j/servers/mcp-neo4j-cypher docker build -t mcp-neo4j-cypher:latest . docker run -p 8000:8000 -e NEO4J_URI='bolt://host.docker.internal:7687' -e NEO4J_USERNAME=neo4j -e NEO4J_PASSWORD='***' mcp-neo4j-cypher:latest ``` ## Intro neo4j-contrib/mcp-neo4j bundles Neo4j MCP servers (Cypher, memory, Aura API); verified 946★ and documents stdio/HTTP/SSE plus timeouts and token limits. **Best for:** Graph-heavy teams who want natural-language Cypher querying and a persistent knowledge-graph memory via MCP **Works with:** Claude Desktop and other MCP clients; Docker for HTTP transport deployments; Neo4j Aura or self-managed Neo4j **Setup time:** 15-35 minutes ### Key facts (verified) - GitHub: 946 stars · 243 forks · pushed 2026-04-10. - License: MIT · owner avatar + repo URL verified via GitHub API. - README-backed entrypoint: `docker run -p 8000:8000 mcp-neo4j-cypher:latest`. ## Main - Treat schema access as the first success metric: the Cypher server exposes schema tools and enforces read timeouts (docs mention a 30s default). - Use token limits to prevent huge query payloads from blowing up context; the Cypher server docs describe truncation using `tiktoken` for read responses. - Run multiple servers with namespacing when you have multiple Neo4j instances (dev/prod or multiple graphs) to avoid tool-name collisions. ### Source-backed notes - Root README lists multiple servers: `mcp-neo4j-cypher`, `mcp-neo4j-memory`, `mcp-neo4j-cloud-aura-api`, and `mcp-neo4j-data-modeling`. - Docs describe transport modes (stdio/SSE/HTTP) and show `--transport http --host ... --port ... --path /api/mcp/` examples. - The Cypher server docs include read timeouts (default 30s) and response token limits for `read_neo4j_cypher`. ### FAQ - **Do I need APOC?**: Yes for schema inspection: the README says APOC must be installed/enabled for schema tools to work. - **Is this the official Neo4j MCP server?**: No — the root README calls these Neo4j Labs servers and links to the official product repo separately. - **How do I keep query results manageable?**: Use read timeouts + token limits as documented in the Cypher server docs. ## Source & Thanks > Source: https://github.com/neo4j-contrib/mcp-neo4j > License: MIT > GitHub stars: 946 · forks: 243 --- ## Quick Use ```bash # Build server image locally (Cypher server subdir): git clone https://github.com/neo4j-contrib/mcp-neo4j.git && cd mcp-neo4j/servers/mcp-neo4j-cypher docker build -t mcp-neo4j-cypher:latest . docker run -p 8000:8000 -e NEO4J_URI='bolt://host.docker.internal:7687' -e NEO4J_USERNAME=neo4j -e NEO4J_PASSWORD='***' mcp-neo4j-cypher:latest ``` ## Intro neo4j-contrib/mcp-neo4j 汇集多个 Neo4j MCP server(Cypher、记忆图、Aura API);已验证 946★,文档覆盖 stdio/HTTP/SSE 与 token limit/timeout 等防护。 **Best for:** 图数据团队:希望通过 MCP 做自然语言 Cypher 查询与可持久化的知识图谱记忆 **Works with:** Claude Desktop 等 MCP 客户端;Docker 部署 HTTP transport;Neo4j Aura 或自建 Neo4j **Setup time:** 15-35 minutes ### Key facts (verified) - GitHub:946 stars · 243 forks;最近更新 2026-04-10。 - 许可证:MIT;作者头像与仓库链接均已通过 GitHub API 复核。 - README 中可对照的入口命令:`docker run -p 8000:8000 mcp-neo4j-cypher:latest`。 ## Main - 先把 schema 取回当作第一里程碑:Cypher server 提供 schema 工具,并支持读取超时(文档提到默认 30 秒)。 - 用 token limit 约束大结果集,避免上下文爆炸;文档说明 read 响应会用 `tiktoken` 按限制截断。 - 多实例场景用 namespacing 同时连接多个 Neo4j(dev/prod 或多图),避免工具名冲突。 ### Source-backed notes - 根 README 列出多个 server:`mcp-neo4j-cypher`、`mcp-neo4j-memory`、`mcp-neo4j-cloud-aura-api`、`mcp-neo4j-data-modeling` 等。 - 文档覆盖 stdio/SSE/HTTP 三种 transport,并给出 `--transport http --host ... --port ... --path /api/mcp/` 示例。 - Cypher server 文档说明 read timeout(默认 30 秒)与 `read_neo4j_cypher` 的 response token limit。 ### FAQ - **需要 APOC 吗?**:需要用于 schema 检查:README 写明 APOC 必须安装并启用才能使用 schema 工具。 - **这是官方 Neo4j MCP server 吗?**:不是。根 README 标注为 Neo4j Labs,并另行链接官方产品仓库。 - **怎么控制返回结果不要过大?**:按文档设置 read timeout + token limit,限制 read 响应尺寸。 ## Source & Thanks > Source: https://github.com/neo4j-contrib/mcp-neo4j > License: MIT > GitHub stars: 946 · forks: 243 --- Source: https://tokrepo.com/en/workflows/neo4j-mcp-servers-cypher-memory-aura-api Author: MCP Hub