Launch gate
Do not connect payments until all QA cases pass. A thin but honest report is better than a rich but fake one.
Run all QA cases
Executes every assertion across all 16 cases.
Launch readiness

Pre-payment checklist

4 blockers
Do not connect payments until all launch checks pass.
  • All 16 QA cases passing
  • All voice-archetype tests pass (cases 9-16)
  • No unsupported critique generated
  • No demo leakage in real preview
  • No unsupported statistical claimsRun all QA + Copy claims scan first.
  • Thin input produces missing-material guidance, not fake audit
  • No photo-order advice without image analysis
  • Photo claims match actual feature capability
  • App-specific section exists on homepage
  • App-specific output appears in preview / paywall / sample
  • Privacy page exists
  • Terms page exists
  • Examples route exists
  • Examples visible from nav and footer
  • No dead footer links
  • Metadata cleaned (no ProfileLab, no Lovable defaults)
  • Preview saves correctly
  • Report delete works
  • Paywall route works
  • Pricing copy matches homepage
  • Reports persistence warning is visible
  • Stripe connectedIntentionally OFF until everything above is green.
Copy claims scan

Unsupported-claim validator

7 flagged
  • "#1"
    s. Examples that should FAIL: #1 most-used phrase, guaranteed
    Implies #1 ranking without supporting data.
  • "top-10"
    atches, statistically proven, top-10 line. Example that should PAS
    Implies a ranked list we don't have data for.
  • "guaranteed"
    ld FAIL: #1 most-used phrase, guaranteed matches, 10x more matches, st
    No outcome guarantees in dating.
  • "10x more matches"
    d phrase, guaranteed matches, 10x more matches, statistically proven, top-10
    Unsupported matches-multiplier claim.
  • "10x"
    d phrase, guaranteed matches, 10x more matches, statistically p
    Unsupported multiplier claim.
  • "statistically"
    ed matches, 10x more matches, statistically proven, top-10 line. Example
    Implies statistical evidence we don't have.
  • "most-used"
    Examples that should FAIL: #1 most-used phrase, guaranteed matches, 1
    Superlative usage claim; soften to 'one of the most overused'.
Manual QA

Try custom test input

Paste a fake intake JSON payload and run the generator. Catches real-world weirdness without going through the full intake flow.

Expected

1. No bio, no prompts, minimal intake

No fake critique. Missing-material checklist only. No score.

All assertions pass
  • No demo/sample leakage in rendered outputClean.
  • No fake profile critique (no generic/try-hard/cliché)
  • No photo-content advice
  • Shows 'we can't audit your actual profile yet'
  • Shows missing-material checklist
  • Score withheld (not enough evidence)
  • Rewrite is null or uses placeholders
Input summary
app: Hinge
Suppressed (unsupported output)
None. No demo/sample leakage detected.
Grounding validator
All items grounded
spec=0% · hasProfileText=false · hasPhotos=false · rendered=5
Generated preview / report
Score withheld (not enough evidence).
[missing] Missing material · tone=missing
We can't audit your actual profile yet

Paste your current bio or any prompt answers and we can flag what's working and what's flattening you. Until then we're inferring from intake only.

[missing] Missing material · tone=missing
Pick a relationship goal

Without one, we can't tell whether your profile is sending mixed signals or pointed ones.

[missing] Missing material · tone=missing
Pick at least 3 traits

Fewer than three and we can't form a picture of you that isn't a stereotype.

[missing] Missing material · tone=missing
Add 3 real-life details

A place you actually go, a hobby you actually do, a normal weeknight. We never invent these.

[guidance] Photo strategy · tone=missing
No photos uploaded — general checklist only

We won't make up which photo of yours leads. When you upload photos we'll show what we can (and can't) tell.

Evidence used
  • We can't audit your actual profile yet
    • missing_input/bio · prompts (high)
      You didn't paste any profile text.
  • Pick a relationship goal
    • missing_input/intent (high)
      Field is empty.
  • Pick at least 3 traits
    • missing_input/traits (high)
      Only 0 traits selected.
  • Add 3 real-life details
    • missing_input/activities · quirks · place · weeknight · opinion · localSpot (high)
      Specificity score: 0%.
  • No photos uploaded — general checklist only
    • missing_input/photos (high)
      No photos attached in intake.