# fabio-rovai/open-ontologies [Health: Active]

**Category:** 🧠 Knowledge & Memory  
**Repository:** https://github.com/fabio-rovai/open-ontologies  
**GitHub Stars:** 1738  
**Views:** 2  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/fabio-rovai-open-ontologies

## Description
AI-native ontology engineering with 39 tools and 5 prompts for OWL/RDF/SPARQL. Validate, query, diff, lint, version, and govern knowledge graphs via Oxigraph triple store.

## Claude Desktop Quick Installation
No confirmed install details yet, so no config is published. Follow the setup steps in the repository above.

## Documentation

## What fabio-rovai/open-ontologies does

fabio-rovai/open-ontologies brings ontology engineering operations into MCP clients. It works with OWL and RDF knowledge graphs and supports SPARQL-oriented querying through an Oxigraph triple store. The project provides 39 tools and five prompts for inspecting, changing, checking, and governing ontology data.

Its focus is change analysis rather than visual ontology editing. A proposed change can be evaluated for added and removed classes, affected triples, risk, conservativity, and newly derived consequences. This makes it suitable for reviewing a semantic change before applying it to a production graph. Protégé remains the appropriate tool for drawing class hierarchies; Open Ontologies evaluates the change around that workflow.

## How it works

The engine is written in Rust and distributed as one binary. It communicates with AI applications through MCP, including Claude and Cursor. Ontology content is processed through an Oxigraph-backed triple store, where the server can validate data, execute queries, compare versions, and apply reasoning rules.

A batch interface can read commands such as `load base.ttl` and `plan proposed.ttl`. Planning compares the proposed ontology with the loaded base and reports semantic effects that a text diff would not reveal. For example, a change that adds one domain assertion can leave the structural blast radius at zero while still producing hundreds of new inferred consequences under the OWL-RL rule table.

The project also supports proof-oriented review. A run can write asserted triples and derivations to files, while `oo-cert asserted.tsv derivations.tsv` checks the resulting certificate. The checker accepts valid derivations and rejects forged conclusions, identifying the rule that fails. Verification can be performed later without an installation of the engine or a network connection.

## Setup and configuration

Run the distributed Open Ontologies binary and connect it to an MCP client that supports the protocol. The project is designed to operate without a JVM or Protégé. Claude and Cursor are named client targets, and the MCP interface is intended for other compatible clients as well.

For command-line workflows, the `batch` command accepts input from standard input. A typical planning flow loads a base Turtle file and then plans a proposed file:

```text
load base.ttl
plan proposed.ttl
```

Certificate verification uses the separate `oo-cert` command with the asserted-triple and derivation files as arguments. The engine and checker are MIT-licensed.

## Tools and capabilities

- Validate OWL and RDF ontology content.
- Query graph data with SPARQL-oriented operations.
- Compare ontology versions and inspect semantic diffs.
- Lint ontology structure and identify modeling problems.
- Plan proposed changes and calculate their consequences.
- Apply rule-based reasoning and report inferred triples.
- Create and independently verify derivation certificates.
- Support ontology versioning and governance workflows through MCP tools and prompts.

The project’s examples include checks for undeclared class terms, invalid `rdfs:range` usage, near-duplicate names, satisfiability, and rejected forged derivations. These checks help distinguish a syntactically small change from one that alters the graph’s inferred meaning.

_Full upstream README: https://allmcps.com/mcp/fabio-rovai-open-ontologies/readme_