Neo4j

AddisDB connects to Neo4j over the Bolt protocol on port 7687 and runs Cypher. Node labels are listed in the sidebar the way tables are for a SQL database, and each result row is rendered into the grid.

Who it is for

Graph databases make relationships first-class. In Neo4j, "friends of friends who bought this" is a short traversal instead of a stack of self-joins, and the cost of following a relationship does not grow with the size of the database.

That last claim has a mechanism behind it: relationships are stored as direct pointers between records, so a hop is a dereference rather than an index lookup. A four-hop query over a hundred million nodes costs about what it costs over a hundred thousand — which is exactly the query a relational schema struggles with.

It suits recommendations, fraud rings, identity and access graphs, network and dependency topology, knowledge graphs, and supply chains — anywhere the connections carry as much meaning as the records.

It is the wrong tool for bulk aggregation over every node of a type; that is a scan, and a columnar engine will win.

Set up the database

1. Neo4j AuraDB (managed): create an instance and download the credentials file when it is offered — the generated password is shown exactly once.
2. Note the connection URI. Aura uses neo4j+s:// with TLS on port 7687.
3. Local: Neo4j Desktop, or docker run -p 7474:7474 -p 7687:7687 -e NEO4J_AUTH=neo4j/password neo4j.
4. On first login to a self-hosted server you will be asked to change the default neo4j password.
5. Create an index on any property you will match on — an unindexed MATCH scans every node with that label.

# A local server: browser on 7474, Bolt on 7687
docker run -p 7474:7474 -p 7687:7687 \
  -e NEO4J_AUTH=neo4j/a-strong-password -d neo4j:5

Connect from AddisDB

1. New Connection → Neo4j under Graph. Port prefills to 7687, database to neo4j and username to neo4j.
2. Enter the host from your AuraDB URI (without the neo4j+s:// scheme) and your password.
3. Set SSL mode to require for AuraDB.
4. Test, then Save.

Port 7474 is the HTTP browser, 7687 is Bolt. AddisDB speaks Bolt — a connection pointed at 7474 will not complete the handshake.

How to query it

MATCH (c:Customer)-[:PLACED]->(o:Order)
WHERE o.total > 500
RETURN c.name AS customer, count(o) AS orders, sum(o.total) AS revenue
ORDER BY revenue DESC
LIMIT 20

Return scalar values rather than whole nodes when you want a readable grid — RETURN c gives you a node object per row, while RETURN c.name, c.email gives you columns. The JSON tab always has the full object if you need it.

Variable-length patterns are where a graph query earns its keep, and also where one can run away from you. Bound the depth: [:KNOWS*1..3] is a traversal, [:KNOWS*] is an invitation to walk the entire graph.

-- Who is within three hops, and how far away
MATCH path = (a:Person {email: $email})-[:KNOWS*1..3]-(b:Person)
RETURN b.name, length(path) AS hops
ORDER BY hops
LIMIT 25

What AddisDB gives you

• Node labels listed in the sidebar so you can see what is in an unfamiliar graph.
• Cypher results flattened into the grid, with the JSON tab for full node and relationship objects.
• Read-only enforcement written for Cypher — CREATE, MERGE, DELETE, SET, REMOVE and DROP are blocked on a read-only connection.
• The Chart view for aggregations, plus AI chat that writes Cypher against your labels.
• The Live Monitor over SHOW TRANSACTIONS, with per-label node counts read from the count store.