Website architecture

The public website should combine narrative documentation with a read-only explorer for published corpus records.

Recommended surfaces

1. Home — what Penrose is and is not.
2. Documentation — the Markdown tree in this directory.
3. Corpus explorer — filters for verdict, family, source, epoch, and reproduction class.
4. Record detail — claim, rationale, provenance, safe metrics, receipt hash, and version lineage.
5. Methodology — gate stack, search accounting, confirmation, and calibration.
6. Status — engine version, active public epoch, calibration date, and publication health.

Rendering strategy

Render documentation statically at build time. Generate corpus index pages or a bounded JSON snapshot for small corpora; move to a read-only API and database only when search volume or corpus size requires it.

In-app reuse

The dashboard should load the same Markdown and navigation.json, using a route mapping such as:

public: /docs/dashboard/verdict-archive
app:    #guide/dashboard/verdict-archive
source: documentation/dashboard/verdict-archive.md

Search

For the initial site, build a client-side search index from public Markdown and bounded record metadata. Do not ship entire private audit logs or source text into the browser search bundle.

Versioning

Publish documentation and corpus snapshots with an engine commit. Stable URLs should resolve to the latest version while immutable version URLs remain available for historical records.