GraphQLite
Graph problems do not always justify a graph database. Standing up and operating a dedicated server is a lot of overhead when what you really have is one application that needs to model relationships well. GraphQLite closes that gap by adding graph capabilities directly to SQLite; you get a single-file, zero-configuration, server-less database that also speaks Cypher.
It supports the core of the Cypher query language, including MATCH, CREATE, MERGE, SET, DELETE, WITH, UNWIND, and RETURN, so you write traversals the same way you would against a full graph engine. A set of graph algorithms is built in, among them PageRank, Louvain community detection, Dijkstra shortest paths, breadth- and depth-first search, and connected components. You can reach it from Python, from Rust, or straight through SQL, and because it is SQLite underneath, it drops into whatever you already run.
Conformance here is measured rather than asserted. GraphQLite runs against the official openCypher Technology Compatibility Kit and currently passes 97.7% of its 3,876 scenarios; the gaps that remain are tracked in the open and concentrate in a few deep areas like timezone-aware arithmetic and nested existential subqueries.
GraphQLite is open source under the MIT license.