Neo4j MCP (Labs)
Four Neo4j MCP servers maintained by the Neo4j Field GenAI team as part of Neo4j Labs: cypher (natural-language-to-Cypher with schema introspection), memory (knowledge-graph memory across agent sessions), cloud-aura-api (Aura cloud instance management), and data-modeling (graph data modelling and visualization). Each is a separate installable PyPI package.
“The Neo4j Labs MCP repo is structurally a four-pack: cypher (natural language to Cypher queries against any Neo4j instance), memory (knowledge-graph storage that persists across agent sessions), cloud-aura-api (manage Neo4j Aura cloud instances), and data-modeling (interactive graph data modelling with Arrows.app import/export). Each is a separate installable PyPI package, so operators can pick the surface they need. The cypher server is the most-used and pairs well with knowledge-graph workflows. The memory server is the editorial standout: persistent knowledge graphs across agent sessions is a primitive unique to Neo4j MCPs in the directory. The honest disclaimer comes from Neo4j Labs themselves: this is an experimental program with active development that sits outside the formal SLAs and backwards-compatibility commitments of the core Neo4j product. For teams who want the official-product Neo4j MCP with vendor support, the README points to neo4j/mcp; the Labs version is the surface for everything else.”
INSTALL THIS SERVER
{
"mcpServers": {
"neo4j-cypher": {
"command": "uvx",
"args": [
"mcp-neo4j-cypher@latest"
],
"env": {
"NEO4J_URI": "bolt://localhost:7687",
"NEO4J_USERNAME": "neo4j",
"NEO4J_PASSWORD": "your-password"
}
}
}
}
{
"mcpServers": {
"neo4j-cypher": {
"command": "uvx",
"args": [
"mcp-neo4j-cypher@latest"
],
"env": {
"NEO4J_URI": "bolt://localhost:7687",
"NEO4J_USERNAME": "neo4j",
"NEO4J_PASSWORD": "your-password"
}
}
}
}
{
"mcpServers": {
"neo4j-cypher": {
"command": "uvx",
"args": [
"mcp-neo4j-cypher@latest"
],
"env": {
"NEO4J_URI": "bolt://localhost:7687",
"NEO4J_USERNAME": "neo4j",
"NEO4J_PASSWORD": "your-password"
}
}
}
}
{
"mcpServers": {
"neo4j-cypher": {
"command": "uvx",
"args": [
"mcp-neo4j-cypher@latest"
],
"env": {
"NEO4J_URI": "bolt://localhost:7687",
"NEO4J_USERNAME": "neo4j",
"NEO4J_PASSWORD": "your-password"
}
}
}
}
{
"mcpServers": {
"neo4j-cypher": {
"command": "uvx",
"args": [
"mcp-neo4j-cypher@latest"
],
"env": {
"NEO4J_URI": "bolt://localhost:7687",
"NEO4J_USERNAME": "neo4j",
"NEO4J_PASSWORD": "your-password"
}
}
}
}
8 TOOLS AVAILABLE
OUR ASSESSMENT
- Four distinct servers covering different parts of the Neo4j workflow; pick the surface you need.
- mcp-neo4j-memory provides persistent knowledge-graph memory across agent sessions, a primitive unique to Neo4j MCPs in the directory.
- Triple transport support (stdio, SSE, HTTP) with HTTP designed for production deployments and load balancing.
- Active development and contributions welcome, with cloud deployment guides for AWS ECS Fargate and Azure Container Apps.
- Neo4j Labs program carries an explicit disclaimer about SLAs and backwards-compatibility; teams needing vendor support should evaluate the official neo4j/mcp.
- mcp-neo4j-cypher requires the APOC plugin installed and enabled on the target Neo4j instance for schema inspection.
- The four-server split adds operational complexity; teams using multiple surfaces install and configure each separately.
Each server connects to Neo4j or Neo4j Aura via standard Neo4j authentication (username/password or OAuth for Aura). Credentials live in environment variables. The cypher server executes generated read and write Cypher queries against the configured database, which means the connecting credentials' permissions bound the agent's blast radius. For production, scope a dedicated Neo4j role with least-privilege grants. The Aura API server holds Aura account credentials and can create, scale, and destroy cloud instances, so credential scope here matters more than for the cypher server alone.
Teams running Neo4j who want either knowledge-graph-backed memory for agents (mcp-neo4j-memory) or natural-language Cypher queries against an existing graph (mcp-neo4j-cypher).
TECHNICAL DETAILS
ADOPTION METRICS
// Reading this940 stars on the multi-server monorepo. Traction is concentrated in the cypher and memory servers; the data-modeling server is newer and growing.
// Reading thisTop community-maintained graph database MCP. The four-server split lets operators install only the surface they need.
SOURCES & VERIFICATION
We don't take any single directory's word for it. Before scoring, we cross-reference 5 public MCP sources, install the server ourselves against the clients we cover, and record when we last re-verified.
The same server, 5 different lenses. We reconcile these signals into our editorial score, which is why our number sometimes diverges from a directory-aggregate star count.
| Source | Their rating | Their star count | Their downloads | Last synced |
|---|---|---|---|---|
| AutomationSwitch This page | 4editorial | 940 | — | APR 28, 2026 |
| PulseMCP | — unrated | unavailable | unavailable | APR 28, 2026 |
| Official MCP Registry | — unrated | unavailable | unavailable | APR 28, 2026 |
| Awesome MCP Servers | — unrated | unavailable | unavailable | APR 28, 2026 |
| Smithery | — unrated | unavailable | unavailable | APR 28, 2026 |
| Glama | — unrated | unavailable | unavailable | APR 28, 2026 |
// Counts are directory-reported; we don't adjust them. Discrepancies usually come from different snapshot times or star-caching.
OTHER DATABASE MCP SERVERS
Postgres MCP
Official MCP server for PostgreSQL. Provides read and write access to PostgreSQL databases with schema introspection, query execution, and transaction support. The reference implementation for database MCP servers.
Supabase
Persistence layer with full Postgres access and Row Level Security awareness. Query tables, manage schemas, handle auth users, and work with storage buckets.
MCP Toolbox for Databases
Google's open-source MCP server for databases. One MCP binary speaks PostgreSQL, MySQL, BigQuery, Spanner, AlloyDB, Cloud SQL, and SQLite. 14,937 GitHub stars and 100 commits on main in the last 30 days. Highest commit cadence in the directory.
ClickHouse MCP
Official ClickHouse MCP server with both ClickHouse Tools (production cluster queries) and chDB Tools (embedded ClickHouse for local analytics). Apache-2.0 with destructive-operation protection: write access is opt-in via configuration.
Redis MCP
Official Redis MCP server with natural-language interface for managing and searching Redis data. EntraID authentication for Azure Managed Redis included; Redis ACL configurable via environment variables. Tool coverage spans strings, lists, hashes, sets, sorted sets, JSON, vectors, and search.
Elasticsearch MCP
Elasticsearch and OpenSearch MCP server with index management, document CRUD, search queries, cluster health, and aggregations. 270 stars and 22 commits on main in the last 30 days.
DISCUSS YOUR
MCP REQUIREMENTS.
Evaluating a server, scoping an internal deployment, or working out whether MCP is the right fit at all. Start the conversation and we will point you at the right piece of the ecosystem.