Repository Scanners
Repository discovery locates supported files and dispatches specialised collectors. It is transport and provenance, not a single collector that understands every file or every RDF resource kind.
What it documents
Family builds discover RDF files supporting ontologies, families and modules,
SHACL,
SKOS, and EKGF use cases.
RDF types and explicit membership determine recognition. A folder named
taxonomies or modules does not establish semantic membership.
Choose local edits or a remote revision
For an existing local checkout with source-owned publication hints:
grapharch publication build-family --local-repository ../my-ontologies \
--output-root output/local-family
This reads the working tree, including uncommitted edits. The documentation KG records that state; it does not claim those edits equal a committed revision.
For a remote GitHub repository:
grapharch publication build-family --github example/my-ontologies \
--git-ref develop --output-root output/remote-family
Replace the repository with your own. A branch tracks its latest fetched commit
on each build; a full commit ID pins an immutable input. --git-repository URL
also accepts other Git hosts. Without --git-ref, the remote default branch is
used. A cached bare repository avoids repeated full clones; jobs read the one
resolved commit, not a changing checkout.
The source must supply unambiguous project/target hints, or the command needs explicit definitions, project, and target overrides. See the family workflow. Family discovery does not automatically import every Markdown, image, code file, or HTML template: those require the appropriate workflow and specialised jobs. Configured Git Markdown, code, image, and legal collectors read pinned blobs with commit and blob provenance. See Markdown and images and code examples for those separate paths.
Collect one RDF file from Git
A website project can explicitly collect one RDF file from a remote repository without discovering and documenting every vocabulary in that repository:
@prefix ga: <https://ekgf.org/ontology/grapharch#> .
<urn:example:docs> ga:collector <urn:example:governance-source> .
<urn:example:governance-source> a ga:CollectorUse ;
ga:collectorType ga:GitRdfFileCollector ;
ga:source "https://github.com/EKGF/ekg-ontologies.git" ;
ga:sourcePath "ekgf-artifact-governance.ttl" ;
ga:gitRef "HEAD" .
This configuration fragment belongs in the surrounding project's publication definitions. It uses the configured-source path of the website build, not a separate new CLI collection kind.
ga:source names the repository. ga:sourcePath selects one regular RDF blob
using a relative path without traversal. ga:gitRef is required: use HEAD
for the latest fetched default-branch commit, a branch or tag name, or a full
commit ID for an immutable selection. The collected graph retains the concrete
observed commit and blob provenance even when the requested reference can move.
Cached Git data avoids cloning the whole repository again on each run.
The collector retains RDF assertions; it does not rewrite imported terms into GraphArch's namespace. Documentation scope and profiles select the terms and arrange their reference entries. To read a publisher's RDF download instead, use the RDF URL Collector.