VizCritique Pro Desktop
Score a Tableau dashboard against a locked 7-domain rubric plus ethics gatekeeper. Read-only. Two-tier delivery: short chat summary + opt-in longform file to Downloads.
When to Use
- You just built a dashboard and want a professional-grade critique before stakeholder review.
- You inherited a dashboard and want to know if it's ready to ship.
- You're prepping for a design-review meeting and want a scored critique to bring in.
When Not to Use
- Rebuilding the dashboard. Use
viz-plannerfor redesign,worksheet-builderfor chart authoring. - Chart-type-only advisory on one worksheet. Use
chart-swap-advisor. - Pre-publish assembly hygiene without a full critique. Use
dashboard-preflight. - Format-string drift. Use
format-drift-scanner. - Text-copy correctness. Use
desktop-spellcheck. - Localization fragility. Use
i18n-scanner. - Data-quality issues. Use
data-quality-review. - Interaction runtime behavior. Use
interaction-validator.
How It Works
Image required. Paste a screenshot of the dashboard. Skill asks if missing. If your dashboard scrolls beyond one screen (aspect ratio 2:1+), paste the top half and the skill will offer per-section re-runs.
Workbook open (optional). If your workbook is open, the skill pulls dashboard structure via get-dashboard-xml to sharpen chart-type and layout scoring. If closed, the skill runs image-only.
Ethics first. Every run starts with an ethics-and-privacy gatekeeper. Critical violations cap the score at 59/100. Warning violations cap at 79/100. Info observations don't cap.
7 domains, locked weights.
| Domain | Weight |
|---|---|
| 1. Audience Adaptation | 12% |
| 2. Message Alignment | 12% |
| 3. Chart Usage | 18% |
| 4. Layout and Hierarchy | 18% |
| 5. Color and Typography | 12% |
| 6. Accessibility | 14% |
| 7. Interactive-Design Signals | 14% |
Two-tier delivery.
- Tier A (chat): Overall score, tier, top 2 strengths, top 3 recommendations, per-domain scores, next-step routing. Under 15 lines.
- Tier B (optional file): Full longform review to your Downloads folder as
.mdor.txt. Executive summary, ethics section, per-domain rationale with signals, rebuild handoff payload.
Tier Table
| Score | Tier | What It Means |
|---|---|---|
| 90-100 | Exemplary | Best-in-class. Publishable as a template. |
| 80-89 | Strong | Meets professional bar. Minor sharpening. |
| 70-79 | Solid | Functional and defensible. Some gaps. |
| 60-69 | Developing | Ships with reservations. |
| 50-59 | Needs work | Not ready for stakeholder review. |
| 0-49 | Rework | Redesign recommended. Route to viz-planner. |
Example Trigger Phrases
- "Review this dashboard"
- "Critique this viz"
- "Score this dashboard"
- "Grade this dashboard"
- "Evaluate my dashboard"
- "How does this dashboard rate"
- "Viz critique"
What the Skill Does Not Score
- Runtime behavior. Design signals only; runtime routes to
interaction-validator. - Data content quality. Underlying data quality routes to
data-quality-review.
Output Handoff
Emits viz_critique_report/1 per the locked schema. Consumed by viz-planner (redesign input at Rework tier), worksheet-builder and chart-swap-advisor (per-chart rebuild payloads), dashboard-preflight (layout findings), format-drift-scanner (color and typography drift), interaction-validator (runtime verification), and workbook-auditor (critique score deltas across runs).
Known Limitations
- One dashboard per run. Batching is refused.
- Image reader reliability on long-form dashboards. Aspect ratios taller than 2:1 trigger per-section re-runs; Layout and Interactive-Design domains flag UNSCANNED on partial coverage.
- Cannot verify interaction runtime behavior.
- Cannot evaluate pixel-level palette compliance beyond accessibility scoring.
- Cannot audit workbook data content.
Distribution
Ships with Tableau Studio. Also available on Skills Exchange for latest-version pulls between Studio releases.