| type | explanation | |||
|---|---|---|---|---|
| status | draft | |||
| owner | project-maintainers | |||
| created | 2026-08-07 | |||
| updated | 2026-08-07 | |||
| last_reviewed | 2026-08-07 | |||
| review_cycle_days | 180 | |||
| summary | Answers to common questions about Ontotect scope, commands, semantics, validation, sources, installation, and licensing. | |||
| canonical | docs/en/faq.md | |||
| related |
|
|||
| supersedes | ||||
| superseded_by |
A suite of 20 independently discoverable Agent Skills generated from one systematic ontology-engineering workflow for design, construction, review, repair, optimization, refactoring, validation, governance, and release. It is not an ontology editor, triple store, or complete reasoner.
Run help, then describe the goal to router. Follow Getting started for a complete first route.
No. router is canonical and route is an alias. Output should normalize to router.
The portable commands are Agent prompt contracts. Use Use Ontotect. Command: review. Target: .... The Python ontotect.py navigator only prints command guidance; it does not perform the ontology work.
No. The npm executable lists, plans, or installs the focused suite into fixed
host discovery roots. It copies the canonical root and generates self-contained
focused skills; for Kilo and OpenCode it also generates thin slash-command
adapters. It does not run the Agent workflow, validators, reasoners, repairs, or
publication. Public 0.1.2 and later include this discovery repair; 0.1.1
remains the historical single-skill installer.
validate is a mode for requested checks. verify is the integrated lifecycle stage that assembles all applicable evidence before a gate. Similarly, build spans a lifecycle while implement creates one vertical slice.
No. SHACL evaluates a data graph against shapes under a configuration. It does not by itself prove OWL consistency, satisfiability, intended entailments, domain correctness, or user acceptance.
Only when an appropriate reasoner is actually run against the intended import closure and configuration. The bundled audit script cannot make that claim and must mark it unverified when no reasoner evidence exists.
No. It compares asserted RDF graphs. A preservation claim also needs inferred hierarchy and entailments, CQs, SHACL, mappings, materialization contracts, and relevant operational results.
No. Use the weakest stack that meets the contract: SKOS for concept schemes, RDF/RDFS for lightweight exchange and inference, OWL for logical entailment, SHACL for graph constraints, and SPARQL for executable information needs. A project may combine them with explicit responsibilities.
No. It distributes original synthesis and public citations, not private PDFs or extracted full text. The evidence register documents coverage and limitations.
Read References and source attribution for standard-form citations and sources.md for the complete evidence map, coverage notes, and authority boundaries.
Yes. Ontotect's original repository content is released under the MIT License. Third-party references and linked works retain their own licenses, and the private research corpus is not distributed under Ontotect's license.
Ordinary ontology work needs preserved inputs, VCS history, graph-aware diffs, recorded tool configuration, and proportional evidence. Stronger integrity controls are appropriate only when a concrete regulated, supply-chain, forensic, or acceptance requirement calls for them.
Follow SECURITY.md. Do not disclose secrets, private ontology data, raw corpus material, or usable exploit details in public issues.