# Wiki feature-driven mockup prompts

These eight prompts were written to explore mockups grounded in the wiki's
*features* — note lifecycle, redlinks, provenance, tags, questions, note
types, the link graph, and embedded diagrams — rather than in its existing
visual design. Each was run through two different image models (ChatGPT and
Codex), producing 16 images total. See
[`2026-07-21-feature-mockups.md`](2026-07-21-feature-mockups.md) for the
assessment of which ideas are relevant to the wiki's actual static-rendered
reading UI.

## 1. Lifecycle-first note view

> Design a note-reading view for a personal knowledge wiki where every
> article has an explicit lifecycle state — seed, working, stable, or
> stale — shown as a first-class visual signal, not a tucked-away badge. A
> "working" note should visibly surface its known material gap; a "stale"
> note should show a warning callout naming what's unsafe to trust and why.
> Show how the same layout adapts across a compact seed-stage stub and a
> long stable article with citations.

## 2. Redlink / wanted-pages research queue

> Design a dashboard for a research queue built entirely from unresolved
> wikilinks ("redlinks") found across a wiki's existing notes — each entry
> shows the linking note, the surrounding sentence for context, and how
> many notes point to the same missing page. The goal is to help a
> researcher triage which gaps are worth writing next based on backlink
> pressure, not alphabetical order.

## 3. Provenance and citation trail

> Design a view that shows a claim inside a note traced back to its
> evidence: frontmatter-level source list vs. inline footnote citations vs.
> the actual archived source artifact (PDF/HTML snapshot). Make the
> distinction between "this note is broadly informed by X" and "this
> specific sentence is backed by Y" visually obvious, including a state for
> a claim with no citation yet.

## 4. Tag graph and merge-candidate audit

> Design an interface for auditing a tag vocabulary across hundreds of
> notes: surface near-duplicate tags (e.g. singular/plural, synonym pairs),
> sparse tags used only once, and dense hub tags, with a one-click
> merge/rename action and a preview of which notes would be affected.

## 5. Question-note frontier

> Design a view of open research questions living as first-class pages in a
> wiki — each showing the question, why it matters, what's already known,
> and what evidence would settle it — arranged so a researcher can see
> which questions are closest to resolution vs. still wide open, distinct
> from an ordinary task list.

## 6. Type-aware note composer

> Design a note-creation flow where the author first chooses a note *type*
> (concept, source, entity, question, synthesis, essay, index, meta) and
> the editor changes its required scaffolding accordingly — e.g. a
> synthesis note demands sources before you can mark it stable, an essay
> attributes a position to the author and treats sources as optional.

## 7. Link-graph / cluster explorer

> Design a whole-vault structural view that partitions notes into
> link-neighborhood clusters and highlights the "bridge" notes connecting
> otherwise separate clusters, plus a list of orphaned notes with zero
> backlinks — built for a maintenance pass, not a general graph toy.

## 8. Diagram/chart-embedded note

> Design a long-form analytical note that mixes connected prose, a Mermaid
> flowchart showing a trust boundary or process, and an inline Vega-Lite
> chart of a trend over time — all rendered identically whether the note is
> viewed live or opened as a static file, with the chart's data source
> cited in the surrounding text.