Rendering¶

Rendering transforms compiled knowledge back into human-readable pages — Markdown, JSON-LD, or HTML — that can be published as documentation or fed to other systems.

How it works¶

riverbank render fetches all facts about an entity from the named graph and produces a formatted page:

riverbank render http://example.org/entity/Acme --format markdown --target docs/knowledge/

Output formats¶

Format	Use case
`markdown`	MkDocs sites, Obsidian vaults, GitHub wikis
`jsonld`	Machine-readable export, API responses
`html`	Standalone pages, email reports

Page structure (Markdown)¶

A rendered Markdown page contains:

Title — entity label or IRI
Summary — key properties in a table
Relationships — connected entities with predicates
Citations — source fragments with evidence spans
Provenance footer — compilation timestamp, profile, confidence

Dependency tracking¶

Each rendered page is stored as a pgc:RenderedPage artifact with dependency edges to its source facts:

<http://example.org/rendered/acme> a pgc:RenderedPage ;
    pgc:dependsOn <http://example.org/fact/acme-produces-widgets> ;
    pgc:dependsOn <http://example.org/fact/acme-founded-1995> ;
    pgc:renderedAt "2024-12-01T10:30:00Z"^^xsd:dateTime ;
    pgc:format "markdown" .

When source facts change, the rendered page is stale and can be regenerated.

Integration with MkDocs¶

Rendered pages in docs/knowledge/ are picked up automatically by the mkdocs-gen-files plugin. No manual navigation entries needed.

The `--persist` flag¶

By default, riverbank render writes the pgc:RenderedPage artifact back to the graph. Use --no-persist to render without recording:

riverbank render http://example.org/entity/Acme --no-persist

Bulk rendering¶

Render all entities:

for iri in $(riverbank query "SELECT DISTINCT ?s WHERE { ?s a ?type }" --format csv | tail -n +2); do
  riverbank render "$iri" --format markdown --target docs/knowledge/
done