What problem does it solve? Running graph algorithms normally requires exporting data out of Snowflake into a separate graph database. This Skill guides an AI agent through using the Neo4j Graph Analytics Snowflake Native App so algorithms like PageRank, Louvain, and Node2Vec execute directly inside Snowflake, with results written back as Snowflake tables. ## Core Features & Use Cases - End-to-end workflow guidance: Covers the full explore → prepare projection views → project-compute-write → inspect flow, including the strict view and column casting rules the graph engine requires. - Full algorithm catalog: Exact SQL CALL syntax and parameters for community detection, centrality, pathfinding, similarity, embeddings, and GraphSAGE graph ML procedures. - Privilege and auth setup: Step-by-step role grants for both app-identity mode and execute-as-user mode with programmatic access tokens, secrets, and caller grants. - Use Case: A data engineer wants to detect fraud rings in a Snowflake transactions table. The Skill instructs the agent to create node and relationship views with properly cast NODEID columns, run Louvain via a single CALL, and join results back to source tables for readable account names. ## Quick Start Ask the agent to run the WCC community detection algorithm on your Snowflake tables using Neo4j Graph Analytics and write the component assignments to an output table.
