Introduction
These are the docs for two projects that form one system.
oxgraph is a storage-agnostic, zero-copy graph
and hypergraph topology substrate for Rust. The core is no_std and
unsafe-free, built around one pipeline — bytes → validate → view → traverse
— with no parse step and no heap graph rebuilt from the bytes.
oxcode is code intelligence for AI agents, built on oxgraph. It indexes source with tree-sitter into an OxGraph-native database, resolves references into graph relations, and serves deterministic, PageRank-curated context over a CLI and an MCP server.
How the pieces fit
The dependency runs in one direction: a topology substrate, an embedded engine built on it, and an agent-facing product built on that.
oxgraph oxcode
substrate engine + product
───────── ────────────────
topology · layouts tree-sitter extraction
snapshots · algorithms ──▶ → resolve → store in oxgraph-db
embedded database → PageRank context → CLI + MCPoxcode depends on oxgraph through a published crate (oxgraph = "0.3.2", with
the db, graph, and topology features), not a local path — so the substrate
and the product version and ship independently.
Start here
- Building on the graph substrate? Begin with oxgraph → Getting started, then Concepts for the topology-vs-meaning-vs-storage split.
- Giving a coding agent code context? Begin with oxcode → Getting started. The MCP server is the headline interface.