What problem does it solve? Running graph algorithms at scale normally requires installing and sizing the embedded GDS plugin on a Neo4j server. This Skill guides agents through Aura Graph Analytics (AGA), Neo4j's serverless on-demand GDS compute, so algorithms run in isolated ephemeral sessions billed per minute without touching the live database. ## Core Features & Use Cases - Session lifecycle management: Authenticate with Aura API credentials, estimate memory with sessions.estimate(), pick a SessionMemory tier, and create, reconnect, list, or delete sessions with TTL control. - Three data source modes: Project graphs from a connected AuraDB instance, a self-managed Neo4j deployment, or standalone Pandas/Spark DataFrames via gds.v2.graph.construct(). - Algorithm execution and write-back: Run PageRank, FastRP, Louvain and ML pipelines in mutate/stream/write modes, poll async jobs, stream node properties, and write results back to Neo4j before cleanup. - Use Case: A data scientist needs PageRank and embeddings over a 1M-node Person graph in AuraDB. The Skill walks through credential setup, memory estimation, remote projection with gds.graph.project.remote, algorithm chaining, and write-back, then deletes the session to stop billing. ## Quick Start Ask the agent to create an Aura Graph Analytics session, project a graph from your AuraDB instance, and run PageRank with results written back to the database.
