R1 Review Gate

Last updated 2026-07-09. Run this before the first tester invite. This is the founder's first-user pass: build the private preview, use it for one real decision, and decide what must change before anyone else sees it.

Build The Preview

Start from a clean tree, then prepare the first-invite private package:

npm run r1:preflight

That command runs the manual pieces in order: npm run prepare:preview, the private package audit (npm run audit:preview:private), full verification (npm run verify:full), both-lane package smoke, status refresh, and the R1 blocker gate. When it passes, it writes dist/r1-preflight-check.json as the local preflight receipt. npm run release:status and docs/PROJECT-STATUS.md summarize that receipt afterward.

Deploy the resulting dist/ folder to the private preview host you will actually send. Open the deployed root URL on desktop and phone. Do not use raw repository files as the review target.

First-User Path

Do this as a real user, not as the maintainer who knows where everything lives:

1. Open the preview root and take the two-minute tour.
2. Open Conscious Consuming from the instance path.
3. Pick one real decision you might actually make this week.
4. Try one flagship category: /app/#explore/banking, /app/#explore/ai-assistants, /app/#explore/clothing, or /app/#explore/learning-resources.
5. Change the values controls and confirm the ranking or explanation changes for a reason you can say aloud.
6. Open one result, inspect its evidence trail, and decide whether the source is enough for the claim being made.
7. Try one non-shopping path: Assembly, Workshop, Slate, or the Adoption Kit.
8. Export or write the receipt below before changing the product.

R1 Gate Questions

Answer these before inviting the first 5 testers:

• Did one real decision get clearer? Name the decision and what changed.
• Where did trust rise or break? Check evidence links, scoring language, privacy posture, and source dates.
• Could you explain the score to a skeptical person? If not, patch before inviting.
• Did any sentence sound generated? Mark exact words, not vibes.
• Did the privacy promise hold? Values, notes, and exports should stay on-device unless you choose to share them.
• Did mobile work without special knowledge? A phone is part of the gate, not a later polish pass.
• Where did the path dead-end? A missing next act is a blocker when it appears on the chosen path.

Decision Rule

Use one of three outcomes:

• Pass to first 5: one real decision became clearer, evidence was findable, no stop condition appeared, and the next action was obvious enough.
• Patch first: the product mostly worked, but a trust break, broken route, unclear score, stale source, generated-sounding sentence, or mobile failure would make the first testers teach us about a known defect.
• Stop: a health, finance, or safety-adjacent claim reads as advice; a preview exposes raw internal files; the evidence trail cannot be found; the privacy promise appears false; or the app looks sponsored or pay-to-rank.

If the outcome is Patch first or Stop, fix the smallest cause and rerun this gate. Do not invite around a known trust break.

R1 Receipt

Copy this receipt into your notes before sending any invite:

Preview URL:
Build stamp:
Preflight receipt:
Date:
Device/browser:
Real decision:
Path tested:
Values changed:
Result inspected:
Evidence source checked:
Did it help?
What was missing?
Did you trust it?
Mobile result:
Copy strikes:
Stop conditions:
Decision: Pass to first 5 / Patch first / Stop
First 5 invitees:
Next patch:

The R1 review is done when this receipt exists and the decision is either Pass to first 5 or a specific patch before rerun.