Best for
- Developers on Neo4j Aura who want to manage instances through a conversational assistant
- Teams that need to quickly spin up or clean up test Aura instances
- Users already running Claude Desktop or another MCP client
N
Neo4j Aura Manager MCP Server
Community
Manage Neo4j Aura cloud instances in natural language
- Category
- Database #50 of 58
- Stars
- ★ 988 Very popular
- Transport
- stdio (local process)
- Runtime
- Python
- Credentials
- API key / credential required
- License
- MIT
- Last commit
45FMRS · D
An instance-management MCP server for Neo4j Aura that lets an AI assistant create, look up, scale, and delete instances in natural language, with simple setup (STDIO plus two required secrets). It is maintained by the Neo4j Field GenAI team as a Labs project rather than an officially supported product, so it suits personal development and experimentation but not production-critical paths that need stability and support; pay particular attention to securing the highly privileged API credentials.
Strongest · Maintenance 11/20 Weakest · Security and permissions 7/20
Why each score
- Reliability 8/20
- The static material contains only the server.json manifest and a repository-wide README; no tests, CI workflows, or verifiable execution evidence specific to this Aura manager server are supplied, so the static calibration cap of 12 applies. The manifest declares a stdio transport for pypi package mcp-neo4j-aura-manager 0.4.8 consistently, so an init and tool-list handshake looks plausible at the configuration level. However, the repository README mixes capabilities of all four servers (list instances, create/destroy instances, scale up and down), so it cannot be confirmed which tools actually belong to this subfolder, leaving the declared-tools-versus-real-behaviour mapping unproven. Aura API calls depend on network and credentials, yet error handling, failure modes, retries, and rate limiting are undocumented, so dependency controllability is unknown. Judged as a plausible happy path with thin edge-case and error-handling evidence.
- Security and permissions 7/20
- The manifest explicitly marks NEO4J_AURA_CLIENT_ID and NEO4J_AURA_CLIENT_SECRET as isSecret and required, showing awareness of credential handling, and calls to the third-party Aura API are expected behaviour with no evidence of credential exfiltration or covert callback. The risks are nonetheless substantial: this server can create, destroy, and resize Aura instances, i.e. billable and destructive operations, yet neither the README nor the manifest documents any confirmation step, least-privilege scoping, API permission narrowing, or data-flow disclosure; credentials are injected into the process as plain environment variables with no key-management guidance. No red line is crossed (no malware, no real tokens in examples, no evidence of default unconfirmed destructive operations), so 0-4 is not warranted, but the absence of permission minimization and dangerous-operation confirmation is a clear deduction.
- Maintenance 11/20
- This is a Neo4j Labs project; the README states it is maintained by the Field GenAI team, welcomes community contributions, and ships an MIT license, indicating an actively developed and clearly licensed open-source effort (980 stars and 25 open issues imply usage and feedback). Manifest version 0.4.8 suggests iterative releases. However, the material provides no release notes, commit history, dependency-update policy, or security-response channel for this specific server, and the README itself disclaims SLAs and backwards-compatibility guarantees, leaving governance and versioning gaps. Missing release and security-response evidence is a deduction.
- Documentation 9/20
- The repository README gives an overview of four servers plus transport modes (stdio/SSE/HTTP) and a cloud-deployment guide link, but that is repository-wide documentation that cannot be attributed to this Aura manager subfolder. Server-specific documentation is absent from the supplied material: no complete tool parameter references, no authentication acquisition steps (how to obtain the Aura Client ID/Secret), no permission scope, no instance sizing or billing impact, no quotas or rate limits, and no troubleshooting. The subfolder link in the README is even written as `./servers/mcp-neo4j-cloud-aura-api//` with a double slash. Documentation stays at the conceptual level with many hidden assumptions, a clear deduction.
- Setup experience 10/20
- The manifest gives a stdio transport and pypi identifier mcp-neo4j-aura-manager 0.4.8 with three environment variable names, so package installation plus environment configuration is theoretically sufficient; the main README also includes HTTP transport examples, but those target mcp-neo4j-cypher and cannot be directly reused here. There are no server-specific install commands, mainstream client configuration snippets, credential acquisition path, or verification steps for the Aura manager, and no CI or test evidence, so the static setup cap of 15 applies. The result is a workable but inference-heavy, incompletely documented setup path.
Static review · not runListed 2026-09-10
Read the FMRS scoring method →Fit and risk
What it can accessUses the networkChanges third-party account data
Not for
- Users looking for the officially supported Neo4j product MCP server (use neo4j/mcp)
- Production environments that require SLAs, backwards compatibility commitments, or a stable long-term API
- Users who do not use Neo4j Aura
Required permissions
- Neo4j Aura API credentials (Client ID and Client Secret) that can create, modify, and delete Aura instances
- Full administrative access to the user's Aura account
Risks and side effects
- The Aura API credentials are highly privileged; if leaked, another party could create, scale, or even delete instances, affecting billing and data availability
- The model may misinterpret natural language instructions, leading to accidental instance deletion or unexpected scaling costs
- As a Neo4j Labs experimental project, it may introduce breaking changes or deprecations with no compatibility guarantees
- An exposed Client Secret in logs or client configuration can leak credentials
Setup
Before you start
Runtime：Python
NEO4J_AURA_CLIENT_ID
Neo4j Aura API client ID (secret); obtained when you create API credentials in the Neo4j Aura console.
NEO4J_AURA_CLIENT_SECRET
Neo4j Aura API client secret, paired with the client ID; keep it safe.
Other optional settings (1)
NEO4J_NAMESPACE
Optional namespace prefix for tool names to avoid conflicts with other servers.
- Create API credentials in the Neo4j Aura console to obtain a Client ID and Client Secret.
- Configure the server in your MCP client: run it with uvx mcp-neo4j-aura-manager (or launch the PyPI package mcp-neo4j-aura-manager 0.4.8) using stdio transport.
- Set the NEO4J_AURA_CLIENT_ID and NEO4J_AURA_CLIENT_SECRET environment variables, and optionally NEO4J_NAMESPACE to namespace the tools.
- Restart the client and confirm the server starts successfully.
Check that it works
After connecting the server in your MCP client, check that the Aura instance management tools appear in the tool list; then ask "List my Aura instances" — a returned instance list confirms the credentials are configured correctly.
Troubleshooting
- Verify that NEO4J_AURA_CLIENT_ID and NEO4J_AURA_CLIENT_SECRET are set correctly and have not expired
- Check that the client configuration uses the command and package mcp-neo4j-aura-manager version 0.4.8
- Review the server startup logs to confirm the STDIO connection has no errors
- If multiple Neo4j MCP servers cause tool name collisions, set NEO4J_NAMESPACE to differentiate them
- Confirm the credentials' Aura account has access to the target instances
Things to try
Once connected, you can ask your AI assistant things like:
- List my Aura instances
- Create a new instance named mcp-test for Aura Professional with 4GB and Graph Data Science enabled
- Find the instance named mcp-test
- Scale down the mcp-test instance
Use cases
List the Aura instances in your account
Create a new instance with a given name, type, memory size, and features such as Graph Data Science
Find instances by name
Scale instances up or down
Delete instances that are no longer needed
Supported clients
Claude DesktopPartial support
VS CodePartial support
CursorPartial support
WindsurfPartial support
Gemini CLIPartial support
Listed from the project's documentation, not tested by this site.
Overview
This is one of the MCP servers in the Neo4j Labs project, located in the repository subfolder servers/mcp-neo4j-cloud-aura-api, published on PyPI as mcp-neo4j-aura-manager, version 0.4.8. It connects to the Neo4j Aura console through the Aura API so that MCP-capable clients such as Claude Desktop, VS Code, Cursor, Windsurf, and Gemini CLI can manage Aura instances in natural language: create and destroy instances, find instances by name, scale them up and down, and enable features. It uses STDIO transport and requires the Aura API Client ID and Client Secret as environment variables, both of which are required secrets. It is an experimental Neo4j Labs project developed and maintained by the Neo4j Field GenAI team, is not supported by the Neo4j product team, and comes with no SLA or backwards compatibility guarantees; the official product MCP server lives in a separate repository, neo4j/mcp.
Similar servers
MCP ClickHouse
79 · B
Connect ClickHouse to your AI assistants
ChEMBL MCP Server
75 · B
Unofficial MCP server providing advanced access to the ChEMBL chemical database
MCP Toolbox for Databases
74 · B
Google's official database MCP toolbox — define AI-accessible database tools via config
DBHub
74 · B
Bytebase's official, minimalist two-tool database MCP server
More servers from this repository
Source revision 1c7ad0793b5f Data synced 2026-10-19 Read the FMRS scoring method