agentic-ontology-development-kit

Search ontology registries for existing terms to reuse before defining new ones. Use during Step 2 (Knowledge Exploration) and Step 3 (Knowledge Organization), or whenever the user asks to find existing ontology terms, search OLS, look up IRIs, check if a concept already exists, or reuse external terms—e.g. "find an existing class for fatigue", "search OLS for cancer", "is there an ontology term for this?", "reuse terms from OBO Foundry".

Verify competency questions against an OWL ontology using SPARQL queries with test data. Use to verify every CQ with real query execution. Also use whenever the user asks to test, validate, or verify an ontology against its CQs, run SPARQL verification, check CQ coverage, or asks "does the ontology answer the competency questions?"—e.g. "verify CQs", "test the ontology", "run CQ checks", "validate against competency questions".

Read, edit, and manage OWL ontologies with OWL-MCP (axioms, prefixes, metadata, pitfall scanning). Use whenever the user wants to add or remove axioms, search axioms, add prefixes, set the ontology IRI, inspect ontology metadata, scan for pitfalls, or formalize an ontology in OWL—e.g. "add this axiom", "find axioms for class X", "add prefix", "check for pitfalls". Align with Ontology Builder workflow (formalization step); use absolute paths for OWL files. Never edit OWL files by hand.

Run ROBOT for ontology QC, merge, reason, convert, template, or query. Use whenever the user needs to verify an ontology, validate OWL profile, merge OWL files, run a reasoner, convert OWL/OBO, build terms from CSV+template, or run SPARQL/DL over an ontology—e.g. "run robot verify", "merge these OWL files", "convert to OBO", "template from this CSV". This is the only skill for ROBOT; do not use odk-run for robot.

Run custom Make targets or ODK tools that have no dedicated skill. Use when the user wants a one-off make target (other than test, prepare_release, refresh-imports, update_repo, odkversion) or to run Konclude, jena, relation-graph, jinjanate, OORT, souffle, dicer-cli, yq-mf, or sssom-cli. Do not use for robot, owltools, dosdp-tools, fastobo-validator, riot, sparql, runoak, or sssom—use those skills instead.

Browse and select Ontology Design Patterns (ODPs) from the ODP catalogue for ontology modeling. Use whenever the user needs to choose a modeling pattern, asks "how should I model X?", wants to represent parts, roles, events, time, collections, situations, n-ary relations, or any recurring ontology structure. Also use proactively during Step 3 (Knowledge Organization) and Step 4 (Draft Proposal) of the ontology-builder workflow when identifying candidate axiom patterns. Covers 55+ patterns from ontologydesignpatterns.org and the ODPA repository.

Understand a target ontology repo's structure, contribution workflow, and build setup. Use whenever the user wants to analyze an ontology repository, learn how to contribute to a project, understand repo layout, or inspect CONTRIBUTING/issue/PR conventions—e.g. "analyze this ontology repo", "how do I contribute to this project", "what's the structure of this repo", or when they provide a repo URL. Prefer this skill over ad-hoc inspection when contributing to external ontologies.

Clone an ontology repository into the workspace's projects directory for local contribution. Use whenever the user wants to clone an ontology repo, set up a project locally, or get a copy to work on—e.g. "clone this ontology", "set up this repo locally", "get a copy of this project to work on". All subsequent work (edits, branch, PR) is done from the cloned project directory.

Create a branch, commit, push, and open a pull request from the current project directory. Use whenever the user wants to create a PR for their changes, open a pull request for an issue, or draft a PR description—e.g. "create a PR for this change", "open a pull request for issue

Summarize and triage ontology repo issues for contribution. Use whenever the user wants to review an issue, find good first issues, summarize open issues for an ontology, or understand what an NTR or other issue needs—e.g. "review this issue", "is this a good first issue", "summarize open issues for this ontology", "what does this NTR need". Use for both single-issue review and listing/batching issues.