Skills become
queryable knowledge

OntoCore compiles natural language skills into OWL 2 ontologies. SPARQL queries instead of probabilistic reading. Deterministic intelligence for any model size.

The Determinism Problem

LLMs read skills probabilistically. Same query, different results. Long skill files are expensive for large models and confusing for small ones.

  • Non-deterministic reading — LLMs interpret text differently each time
  • Token waste — Large models burn tokens reading long skill files
  • Small model limits — Complex skills are unreadable by edge models

The OntoCore Solution

Transform skills into formal ontologies. Query with SPARQL. Deterministic answers, entailment reasoning, democratized intelligence.

  • Deterministic queries — SPARQL returns exact answers, every time
  • Entailment reasoning — Infer dependencies, conflicts, capabilities
  • Democratized intelligence — Small models query what large models read

Key Capabilities

Everything you need to build deterministic, queryable skill ontologies.

🧠

LLM Extraction

Claude extracts structured knowledge from SKILL.md files with tool-use.

🏛️

Knowledge Architecture

Follows "A is a B that C" definition pattern (genus + differentia) for precise classification.

📊

OWL 2 Serialization

Outputs valid OWL 2 DL ontologies in RDF/Turtle format with formal semantics.

SHACL Validation

Constitutional gatekeeper ensures logical validity before any write operation.

⚙️

State Machines

Skills define preconditions, postconditions, and failure handlers for robust execution.

🔒

Security Pipeline

Defense-in-depth: Unicode normalization + regex patterns + LLM review for threat detection.

How It Works

From natural language to validated OWL 2 ontology in one pipeline.

OntoCore Architecture

Neural Layer

Claude extracts structured knowledge from natural language using tool-use.

Symbolic Layer

OWL 2 ontology stores knowledge with formal Description Logic semantics.

Query Layer

SPARQL provides precise, indexed retrieval for deterministic results.

Roadmap

This evolves as we build.

Phase 1

OntoCore

Neuro-symbolic core: SKILL.md → OWL 2 DL ontology

Phase 2

OntoSkills

Compiled skill library ready for agent queries

Phase 3

OntoMCP

Rust MCP server with 12 semantic tools

🔨
Phase 4

OntoStore

Centralized marketplace for ontologies

💡
Phase 5

OntoClaw

Autonomous agent powered by structured knowledge

Track progress on GitHub

Stop reading. Start querying.

Compile your skills into validated ontologies. Query with SPARQL. Reason with confidence.