GMEOW

One engine, three products

The v0.2.0 realignment ships the substrate as products: no consumer ever needs to learn RDF to benefit. The three product surfaces are shipped, and the engine below is published through the full validation, reasoning, documentation, and release pipeline. Roadmap: epic #300 and the realignment specification.

PyPI wheels and sdists for gmeow and gmeow-gts are built in GitHub Actions and signed with GitHub artifact attestations; release artifacts also include SPDX SBOMs. After downloading a package, verify it with gh attestation verify <path-to-wheel-or-sdist> --repo Blackcat-Informatics/gmeow-ontology.

Quick start

No source checkout is required to use the public CLI and GTS package; the wheel carries the bundled gmeow.gts snapshot.

pip install gmeow gmeow-gts

gmeow info
gmeow describe gmeow:StandpointClaim
gmeow transpile source.ttl --profiles all -o out/
gmeow docs --directory docs-tree
gmeow mcp

At a glance

What it adds

Modules

The ontology is composed of 62 guided slices covering identity, social/contact/email/account data, content and evidence, software, trust, attestation, cognition, epistemics, agreements, rights, norms, risk, place, time, events, music, narrative, research objects, and the frame-relative observation and measurement spine. A slice is a self-contained unit (module, shapes, alignments, queries, tests, documentation) whose manifest is the sole source of identity and tier; the logic layer is native RDF 1.2 first, with OWL, Datalog, N3, Prolog, probabilistic, counterfactual, and profile-certified forms generated as projections of the same source.

The gmeow.gts bundle

GTS is the verifiable, append-only, content-addressed transport format for single-file agent memory. The public snapshot carries the ontology, claim layer, evidence blobs, projection surface, docs, and verification trail together for repo-free CLI use and offline verification.

The current dist-profile snapshot contains 18,207 terms, 33,142 quads, 116 RDF 1.2 reifiers, 324 statement annotations, and 68 content-addressed blobs.

Alignments

GMEOW does not replace existing vocabularies — it aligns to them: SSSOM, EDOAL, FnO and SPARQL projection artifacts are generated from one mapping DSL, tying GMEOW terms to external standards so it can reason on its own yet interoperate with other datasets.

Alignment targets include: FOAF, schema.org, REL, DOAP, PROV-O, ORG, OntoLex-Lemon, BIO, vCard, BIBO, Wikidata, BFO, GeoNames.

Mappings are maintained as SSSOM tables (mappings/) and compiled into gmeow-alignments.ttl; cross-dataset owl:sameAs links live in gmeow-linksets.ttl.

Downloads

BII publishes the full GMEOW snapshot, not just the human-readable page. The four canonical RDF files live at the site root as content-negotiation targets for /gmeow; profile graphs, GTS packages, closures, alignments, projection examples, term exports, metadata, context, logo, citation metadata, and native docs extracted from GTS live under /gmeow/.

Apache publication layout

BII's Apache config publishes /gmeow as a slash-namespace ontology IRI. Browsers receive this HTML page; clients asking for RDF media types receive 303 See Other to the matching serialization. The rules also set RDF media types, allow CORS on ontology artifacts, and mark the negotiable endpoint with Vary: Accept plus Cache-Control: private, no-store.

Reasoning & validation

GMEOW's logical TBox stays deliberately within OWL 2 DL, so decidable reasoners apply. The pipeline merges the import closure first, then runs OWL 2 DL profile checks, ELK consistency, release-grade HermiT closure, SHACL completeness shapes, entailment-based competency tests, and no-drift checks between the RDF 1.2 / RDF-star statement artifact and its OWL axiom-annotation downcast.

Full term reference

The native RDF 1.2-aware class-by-class, property-by-property reference documentation is in /gmeow/docs-full/reference/. Each term IRI (for example, …/gmeow/Person) 303s to a native term page such as /gmeow/terms/Person/. Source, contribution guide and release history live in the GitHub repository.

The vocabulary is licensed CC BY 4.0 and the tooling Apache-2.0; also available under separate proprietary terms. © 2026 Blackcat Informatics® Inc.