{"id":"neo4j-api","name":"Neo4j","homepage":"https://neo4j.com","repo_url":"https://github.com/neo4j/neo4j","category":"database","subcategories":["graph-database","knowledge-graph","analytics"],"tags":["neo4j","graph-database","cypher","knowledge-graph","aura","bolt-protocol","rest-api","relationships","network-analysis","fraud-detection"],"what_it_does":"Neo4j is the world's leading graph database, storing data as nodes, relationships, and properties rather than rows and columns. It uses the Cypher query language for expressing graph traversals, pattern matching, and aggregations. Neo4j exposes data via the Bolt binary protocol (used by all official drivers), an HTTP API for Cypher execution, and a REST Discovery API. It is available as Community Edition (open source), Enterprise Edition, and Neo4j Aura (fully managed cloud). A suite of four community-maintained MCP servers (from Neo4j Labs) enables AI agents to query, model, and manage graph data via natural language.","use_cases":["Knowledge graph construction and traversal for AI reasoning and context enrichment","Fraud detection using multi-hop relationship traversal across transactional networks","Recommendation engines that use social graph topology (friends-of-friends, collaborative filtering)","Supply chain and network impact analysis using graph pathfinding algorithms (Dijkstra, PageRank)","Entity resolution and data lineage tracking across complex interconnected datasets","Semantic knowledge bases and ontologies powering RAG with structured relationship context"],"not_for":["Relational/tabular data that doesn't have meaningful entity relationships","High-throughput OLAP analytics over billions of rows (use ClickHouse or BigQuery)","Simple key-value or document storage without graph traversal needs","Teams not willing to learn Cypher query language"],"best_when":"Your data is fundamentally relational — fraud networks, knowledge graphs, recommendation graphs, org charts — where traversing relationships is the primary query pattern.","avoid_when":"Your data is tabular with few meaningful relationships, or you need sub-millisecond OLAP aggregations over billions of rows.","alternatives":[{"id":"mcp-neo4j","reason":"MCP-only interface without direct REST/Bolt API access"},{"id":"qdrant-api","reason":"Better for vector similarity search; no graph traversal support"},{"id":"elasticsearch-api","reason":"Better for full-text search and log analytics over flat document collections"}],"af_score":73.8,"security_score":65.0,"reliability_score":null,"package_type":"mcp_server","discovery_source":["github"],"priority":"low","status":"evaluated","version_evaluated":"current","last_evaluated":"2026-03-01T09:50:05.987729+00:00","performance":{"latency_p50_ms":10,"latency_p99_ms":100,"uptime_sla_percent":99.95,"rate_limits":"No published rate limits; bounded by cluster resources","data_source":"llm_estimated","measured_on":null}}