open-ontologies

Open Ontologies is a Rust MCP server and a Tauri desktop app (Studio) for ontology engineering with AI. The README calls it a Terraforming MCP for knowledge graphs: it validates, classifies and governs AI-generated ontologies, ships as a single binary, and needs no JVM or Protégé. Its 48 tools, prefixed `onto_` over MCP and also available as CLI subcommands, let Claude build, validate, query, diff, lint, version, reason over and align ontologies.

The tool table is broad. Core tools handle RDF and OWL validation, loading, querying and conversion. A marketplace installs 30 standard W3C, ISO and industry ontologies. Other groups cover PostgreSQL-to-OWL schema import, structured-data-to-RDF mapping with SHACL, named snapshots and rollback, Terraform-style lifecycle commands (plan, apply, lock, drift, enforce, monitor, lineage), cross-ontology alignment, ICD-10, SNOMED and MeSH crosswalks, dual-space semantic search, and a native OWL2-DL reasoner. Studio adds a 3D force-directed graph, an AI chat panel and a property inspector.

The README publishes benchmarks. On OntoAxiom, an axiom identification test over 9 ontologies and 3,042 axioms, it reports an F1 of 0.717 for MCP extraction against 0.431 for bare Claude Opus and 0.197 for the paper's best o1 result. It also says a one-sentence Pizza ontology prompt produced 95 of 99 reference classes in about five minutes. These are the author's own results. Install uses pre-built binaries for macOS and Linux or `cargo build --release`, and the licence is MIT. Last push was 2026-10-13. This review is based on the README and repository data; we have not run it ourselves.