Neo4J
Graph database platform for connected data and real-time analytics
Verdict
Common use cases
- Prototype Cypher queries during schema design
- Provision test databases for feature branches
- Snapshot production graphs before migrations
- Audit user access across Aura projects
- Estimate GDS memory for algorithm runs
Integration
- Vendor
- Neo4J
- Category
- developer-tools
- Auth
- OAUTH2
- Tools
- 21
- Composio slug
neo4j
Tools
- Aggregate Directors Count
Executes a directorsAggregate GraphQL query to count directors in Neo4j. Use when you need to get the total count of director records in the database. The GraphQL query returns the count of all directors regardless of pagination.
- Create Neo4j Aura Instance
Creates a new Neo4j Aura instance with the specified configuration. Use this action when you need to provision a new Neo4j Aura database instance. The operation is asynchronous - the API returns 202 Accepted while the instance is being crea
- Create Neo4j Aura Snapshot
Create an on-demand snapshot of a Neo4j Aura instance. Triggers asynchronous backup creation and returns a snapshot ID that can be used to monitor progress. Use this action when you need to manually create a backup point before major change
- Estimate GDS Session Size
Estimate the size of a new GDS (Graph Data Science) session based on node/relationship counts and algorithm categories. Use this action when you need to determine the appropriate GDS session size before creating a new graph analytics sessio
- Get Neo4j Aura Instance Details
Retrieves details for a specific Neo4j Aura instance using its unique instance ID. Returns comprehensive information including instance name, status, cloud provider, region, memory allocation, connection URL, and other configuration details
- Get Neo4j Aura Organization User
Retrieves details of a specific user in a Neo4j Aura organization. Returns user information including email, organization roles, MFA status, last activity, and project-level access permissions. Use this action when you need to inspect user
- Get Neo4j Aura Project Details
Retrieve details of a specific Neo4j Aura project (tenant). Returns project information including its ID, name, and available instance configurations with regions, types, memory, storage options, and cloud providers. Use this action when yo
- Get Neo4j Aura Project Details (v1beta5)
Retrieve details of a specific Neo4j Aura project (tenant) using the v1beta5 API. Returns project information including its ID, name, and available instance configurations with regions, types, memory, storage options, and cloud providers. U
- Get Neo4j Aura Snapshot Details
Retrieves details of a specific snapshot for a Neo4j Aura instance. Returns the snapshot profile, current status, and creation timestamp. Use this action when you need to check the status or details of a particular backup snapshot.
- List Aura Projects
List all Aura projects (tenants) available to the authenticated user. Returns a summary of each project including its unique ID and name. Use this action when you need to retrieve all Aura projects to identify which projects are accessible
- List GDS Sessions
List all Graph Data Science (GDS) sessions for the authenticated Neo4j Aura account. Returns a summary of each GDS session including status, memory allocation, host information, and expiry date. Use optional filters (instanceId, tenantId, o
- List IP Filters
Returns a list of IP filters configured for a Neo4j Aura organization. Use this action when you need to retrieve all IP filters to check which CIDR blocks are allowed or blocked, or to determine which resources are protected by IP filtering
- List Neo4j Aura Agents
List all agents belonging to a specific Neo4j Aura project. Returns agent information including ID, name, description, database ID, privacy settings, endpoint links, MCP status, and attached tools. Use this action when you need to retrieve
- List Neo4j Aura Instances
Returns a list of Neo4j Aura instances. Use when you need to retrieve all available Neo4j Aura instances in your account, optionally filtered by tenant.
- List Neo4j Aura Snapshots
Lists available snapshots for a Neo4j Aura instance. Returns both scheduled and ad-hoc snapshots with their status and creation timestamps. Use when you need to retrieve information about available database backups or recovery points for a
- List Project Users
List all users in a specific Neo4j Aura project. Returns user information including email, user ID, and project roles for each user. Use this action when you need to retrieve all users associated with a project to identify team members or v
- Pause Neo4j Aura Instance
Pauses a Neo4j Aura instance. Use when you need to temporarily stop an instance to reduce costs or perform maintenance. The instance can be resumed later using the ResumeInstance action. This action is asynchronous - the instance will be in
- Restore Neo4j Aura Instance from Snapshot
Restore a Neo4j Aura instance from a snapshot. Replaces all current data with the snapshot data. This action is irreversible — all data since the snapshot was taken will be permanently lost. Use this action when you need to recover an Aura
- Update IP Filter
Updates an existing IP filter for a Neo4j Aura organization. Use this action when you need to modify an existing IP filter's settings, such as updating its name, description, allow list entries, or enabling/disabling the filter. The request
- Update Neo4j Aura Instance
Edits the configuration of a Neo4j Aura instance. Allows updating the instance name, memory allocation, storage size, and plugin configurations. Use this action when you need to modify an existing Neo4j Aura instance's settings such as rena
- Update Neo4j Aura Instance
Updates a Neo4j Aura instance configuration. Allows modifying instance properties such as name, memory allocation, storage size, vector optimization, and graph analytics plugin. Use this action when you need to modify an existing Aura insta
Setup
Setup guide
- 11. In Switchy, open your workspace settings and navigate to the Integrations tab. 2. Find Neo4j in the developer tools section and click Connect. 3. You'll be redirected to Neo4j's OAuth2 consent screen — sign in with an account that has admin access to your Aura organization. 4. Grant the requested scopes: these include read access to instance metadata, write access for provisioning and snapshots, and user management permissions. 5. After authorizing, Switchy redirects you back and confirms the connection. 6. Open any Space and type '@Neo4j list my Aura instances' to test — the AI should return a summary of your active databases. 7. To query a specific graph, reference the instance ID in your prompt (find it by asking Neo4j for instance details first). 8. For GraphQL queries, ensure your target instance has a GraphQL endpoint enabled in Aura's console.
What teammates see: by default, memories from Neo4J are scoped to the Project (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Projects have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
List Active Instances
@Neo4j show me all active Aura instances in my organization, including their memory allocation and cloud region
Query Graph Relationships
@Neo4j execute a directorsAggregate query on instance [instance-id] and tell me how many director nodes exist
Create Development Database
@Neo4j create a new Aura instance named 'feature-test-db' in AWS us-east-1 with 2GB memory and default storage
Backup Before Deploy
@Neo4j create a snapshot of instance [instance-id] and confirm when the backup job completes
Size GDS Session
@Neo4j estimate the GDS session size needed for 500k nodes, 2M relationships, and running community detection algorithms
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@neo4j create a new Aura instance named 'ml-experiments' with 4GB memory in us-west-2 for our data science team
Instance creation initiated. The Aura API returned 202 Accepted with instance ID 'a1b2c3d4-5678-90ef'. Provisioning typically takes 3-5 minutes. The instance will be accessible at ml-experiments-a1b2c3d4.databases.neo4j.io once status changes from 'creating' to 'running'. Initial credentials will be sent to your registered email. Estimated cost: $200/month based on 4GB tier pricing.
This write operation is asynchronous — the MCP triggers provisioning but doesn't wait for completion. You'll need billing permissions in addition to instance creation scope. The example shows how Switchy can orchestrate infrastructure changes, but your team should establish approval workflows before letting AI create billable resources. Check instance status separately after a few minutes.
@neo4j estimate the GDS session size I'd need to run PageRank on a graph with 2 million nodes and 8 million relationships
For a graph with 2M nodes and 8M relationships running PageRank (centrality algorithm category), the GDS estimator recommends a session size of approximately 12-16GB. This accounts for the in-memory graph projection (~6GB), algorithm workspace (~4GB), and overhead (~2-6GB). The estimate assumes default PageRank parameters. If you plan to run multiple algorithms concurrently or use write-back operations, add 20-30% buffer. Consider starting with a 16GB session to avoid out-of-memory errors.
This synthesis example pairs Neo4j's GDS sizing API with the AI's ability to interpret scale requirements and translate them into actionable recommendations. The estimate is based on Neo4j's internal heuristics but doesn't account for your specific data density or query patterns. Always test with a representative sample before committing to production session sizes. Requires GDS-enabled Aura tier or self-hosted Enterprise.
Use-case deep-dives
When Neo4j MCP wins for ephemeral graph testing
A 6-engineer team building a recommendation engine needs to test graph algorithms against realistic data volumes before committing to production schemas. The Neo4j MCP is the right call here because it automates Aura instance creation and snapshot management through OAuth2, letting developers spin up isolated test databases in Slack threads without touching the console. The GDS session size estimator tool prevents the common mistake of under-provisioning memory for algorithm runs. This works cleanly for teams running 3-10 experiments per sprint. If your team is testing more than 20 graph variations weekly, the 21-tool surface area becomes noise—script the Aura API directly instead. For teams prototyping their first graph feature, this MCP collapses a 4-step provisioning workflow into a single agent request.
When Neo4j MCP falls short for support ticket routing
A 12-person support team wants to query a knowledge graph of product issues and resolutions during live customer calls. The Neo4j MCP is borderline here. The directorsAggregate and similar GraphQL query tools let agents pull relationship counts and traverse entity links without writing Cypher, which is useful for non-technical support staff. OAuth2 auth means no credential sharing across the team. But the MCP exposes instance management tools (create snapshots, estimate GDS sessions) that support agents will never use, and it lacks read-optimized query builders for common support patterns like 'find all tickets linked to this error code.' If your support graph has under 50k nodes and queries are ad-hoc, the MCP works. Beyond that scale or for repetitive lookup patterns, build a custom FastMCP server with 4-5 support-specific tools instead.
When Neo4j MCP handles ops monitoring at scale
A 3-person platform team manages 18 Neo4j Aura instances across US, EU, and APAC regions for a SaaS product. The Neo4j MCP is the right tool because it centralizes instance detail retrieval and user permission audits in a single OAuth2 session. During incident response, engineers can query memory allocation, connection strings, and snapshot status across all regions from a shared Switchy workspace without switching between Aura consoles. The project details tools surface configuration drift (one region running older memory tiers) that manual checks miss. This setup works for teams managing 10-30 instances. If you're operating 50+ instances or need sub-minute health polling, the MCP's per-request latency becomes a bottleneck—use the Aura API with a monitoring stack instead. For small platform teams doing weekly audits, this MCP turns a 40-minute manual review into a 3-minute agent query.
Frequently asked
What does the Neo4j MCP let me do in Switchy?
It connects your team to Neo4j Aura — Neo4j's managed cloud graph database. You can provision new instances, snapshot databases, query director counts via GraphQL, estimate Graph Data Science session sizes, and manage organization users. It's built for teams running graph workloads who want AI agents to handle routine Aura admin tasks without switching to the Neo4j console.
Do I need Neo4j admin access to connect this MCP?
Yes. The OAuth2 flow requires permissions to create instances, manage snapshots, and read organization users. If you only have read access to a single project, most tools won't work. Connect it with an account that has org-level admin or project owner rights in Neo4j Aura, or you'll hit authorization errors on provisioning and snapshot actions.
Can the Neo4j MCP run Cypher queries against my graph?
No. The 21 tools focus on Aura infrastructure — creating instances, estimating GDS session sizes, pulling project details. For actual Cypher queries or graph traversals, you'd use Neo4j's native drivers or the GraphQL endpoint directly. This MCP is for managing the database platform, not querying your data model.
How is this different from using Neo4j's REST API directly?
The MCP wraps Neo4j Aura's management APIs so your AI agents can provision databases or snapshot backups in natural language. If you're already scripting Aura tasks with curl or the Python client, this won't add much. But if your team wants conversational access to "spin up a dev instance" or "check our GDS estimate", the MCP saves you writing that glue code.
Who on my team should connect the Neo4j MCP?
Whoever owns your Neo4j Aura organization or project billing. Since the MCP can create billable instances and snapshots, you don't want every engineer connecting it. One shared connection in Switchy, scoped to a workspace with your data platform team, keeps spending predictable and audit logs clean.