VibeScored

How we score AI-built software

VibeScored ranks software built with AI coding tools — Lovable, Bolt, v0, Cursor, Claude Code, and Replit — using fully automated scoring across four dimensions: Quality, Traction, Efficiency, and DSGVO compliance. Every score is computed from external APIs. No subjective ratings, no editorial discretion.

If you disagree with a score, file a dispute — we re-run the measurement with full transparency.

01

Quality Index

Technical quality, fully automated

The Quality Index is a 0–100 score that measures the technical quality of an AI-built project. It is computed nightly from four external APIs: Google PageSpeed Insights for performance, Mozilla Observatory for security headers, a custom DSGVO compliance scanner, and axe-core for accessibility.

No self-reporting is involved. Every data point is fetched and verified independently. Projects must serve HTTPS to be eligible for scoring. Having a reachable Impressum and Datenschutz page earns bonus compliance points but is not required for eligibility. See the full Quality leaderboard.

Quality Index formula
30%
Performance
25%
Security
25%
Compliance
20%
Accessibility

Performance (30%)

Google PageSpeed Insights API
  • Lighthouse performance score on mobile strategy
  • Mobile-first: most real users browse on mobile devices
  • Catches AI-built apps that only work on desktop

Security (25%)

Mozilla Observatory API
  • HTTP security headers: HSTS, CSP, X-Frame-Options, X-Content-Type-Options
  • Full score requires A+ rating across all header categories
  • Catches the security headers most AI code generators forget

DSGVO Compliance (25%)

Custom compliance scanner
  • 10 individual DSGVO checks, weighted by legal importance
  • Impressum, Datenschutzerklärung, cookie consent, EU hosting
  • No pre-consent tracking scripts or Google Fonts CDN

Accessibility (20%)

axe-core via Playwright
  • Automated WCAG 2.1 Level AA conformance checks
  • Headless Chromium tests 5 randomly selected pages
  • Results averaged across all tested pages
Extended metrics (when available)

Projects with public GitHub repositories and sufficient Lighthouse data unlock three additional quality signals. When available, these metrics are folded into the Quality Index at 30% combined weight (Bundle 10%, Code Health 12%, AI Smells 8%), with the four core metrics redistributed to 70%.

Bundle & Payload (10%)

Google Lighthouse
  • Total JS bundle size, unused JavaScript ratio, HTTP request count
  • Under 200KB JS = full score, over 2MB = zero
  • Catches bloated AI-generated bundles with unused imports

Code Health (12%)

GitHub API
  • Dependency vulnerabilities via Dependabot alerts
  • TypeScript strict mode, any/ts-ignore usage count
  • Test file presence and test-to-source ratio
  • Linter and formatter configuration detected

AI Code Smell Score (8%)

Custom detector
  • 12 automated checks for AI-generated code patterns
  • Catches: excessive inline styles, missing error boundaries, hardcoded secrets
  • Console.log in production, missing alt text, duplicate libraries
  • Severity-weighted: critical (-20), warning (-8), info (-3)
Quality Index (extended)
Core (70%)
20%
Performance
18%
Security
18%
Compliance
14%
Accessibility
Extended (30%)
12%
Code Health
10%
Bundle & Payload
8%
AI Code Smells
02

Traction Index

Real-world adoption and reliability

The Traction Index is a 0–100 score that measures whether real people use a project and whether it stays online reliably. It combines visitor traffic (via Plausible Analytics), uptime monitoring (via UptimeRobot), and project longevity.

No vanity metrics. Visitor counts use a logarithmic scale so a single viral hit cannot dominate — a project alive for one year with 500 monthly visitors outscores a one-week wonder with 50,000 visitors. Eligibility floor: 90-day uptime must be at least 90%. See the full Traction leaderboard.

Traction Index formula
40%
Visitor Score
30%
Uptime Score
30%
Longevity

Visitor Score formula

Logarithmic scale that rewards sustained usage over viral spikes.

Visitor Score = min(100, 20 × log₁₀(monthly_visitors + 1))
100 visitors/mo40
1,000 visitors/mo60
10,000 visitors/mo80
100,000 visitors/mo100

Because of the log scale, 10,000 visitors/month scores the same as 100,000 would in a linear model. You cannot win on traffic alone.

Uptime score mapping (90-day window)

≥ 99.9%100
99.0 – 99.9%80 – 99
95.0 – 99.0%40 – 79
< 90%Ineligible

Longevity formula

Rewards projects that stay online and maintained over time.

Longevity = min(100, days_alive / 3.65)
30 days online≈ 8 points
180 days online≈ 49 points
365 days online= 100 points (max)
03

Efficiency Index

Value shipped per dollar and per token

The Efficiency Index is a 0–100 score that measures how much value was shipped per dollar spent and per AI token consumed. It is only available for Silver and Gold verified projects — self-reported build stats at Bronze tier do not affect this leaderboard.

This index rewards builders who get more done with fewer resources and less rework. It uses percentile ranking across all verified projects, so scores are relative to the cohort. See the full Efficiency leaderboard.

Efficiency Index formula
50%
Cost Efficiency
30%
Token Efficiency
20%
Iteration Discipline

Cost Efficiency (50%)

cost / feature_count

Lower cost per shipped feature results in a higher percentile rank across all verified projects in the cohort.

Token Efficiency (30%)

tokens / feature_count

Fewer AI tokens consumed per shipped feature means higher rank. Measures how efficiently builders use prompts.

Iteration Discipline (20%)

100 − 15×refactors − 25×abandoned

Penalizes excessive major refactors and abandoned features. Getting it right with fewer iterations matters.

04

Builder Index

Aggregate reputation score across all projects

The Builder Indexis a 0–100 reputation score for individual developers. It averages Quality, Traction, and Efficiency across all of a builder's live projects, then applies bonuses for Gold-verified work and penalties for upheld disputes or confirmed fakes.

Companies hiring from the job board use the Builder Index to find top-ranked developers. See the full Builder leaderboard.

Builder Index formula
Base = 0.40 × avg(Quality) + 0.40 × avg(Traction) + 0.20 × avg(Efficiency)
Bonuses
+5 per Gold-verified project
Maximum bonus: +25 (5 Gold projects)
Penalties
−10 per upheld dispute (12-month window)
−25 per confirmed fake (12-month window)
Final Builder Index = clamp(0, 100, Base + Bonus + Penalty)
05

Composite Score

The headline number on every project card

The Composite Score is the single 0–100 number displayed on project cards, leaderboard rows, and embeddable badges. It combines Quality, Traction, and (when available) Efficiency into one headline metric.

When Efficiency data is unavailable (Bronze tier projects), the formula automatically falls back to a 60/40 Quality/Traction split, ensuring fair comparison across verification levels.

With Efficiency data (Silver / Gold)

Three-factor formula

0.40 × Quality Index
0.40 × Traction Index
0.20 × Efficiency Index
Without Efficiency data (Bronze)

Two-factor fallback

0.60 × Quality Index
0.40 × Traction Index
06

Verification tiers for AI-built projects

Trust levels for build data

VibeScoreduses three verification tiers — Bronze, Silver, and Gold — to indicate how trustworthy a project's self-reported build data (hours, cost, tokens) is. Higher tiers unlock the Efficiency leaderboard and earn Builder Index bonuses.

All projects start at Bronze. Submit your project to get started, then upload chat logs or connect your AI tool's API to upgrade.

Bronze
Self-declared
  • Self-reported build stats
  • Quality + Traction scoring
  • Embeddable badge
  • Efficiency leaderboard
  • Builder Index bonus
Silver
Log-verified
  • Chat log upload reviewed by team
  • Build stats cryptographically attested
  • Efficiency leaderboard eligible
  • Quality + Traction scoring
  • Builder Index bonus
Gold
API-verified
  • Direct API integration with AI tool
  • Real-time build data feed
  • Efficiency leaderboard eligible
  • +5 Builder Index bonus per project
  • Highest trust badge on profile
07

DSGVO compliance scoring for AI-built software

10 automated checks, weighted by legal severity

VibeScoredruns a custom automated DSGVO (GDPR) compliance scanner against every indexed project. Each of the 10 checks is weighted by its legal importance in the German, Austrian, and Swiss (DACH) market. This is the platform's key differentiator — no other AI software leaderboard scores DSGVO compliance automatically.

The compliance score feeds into the Quality Index at 25% weight, making it impossible to rank highly without meeting basic European data protection requirements.

HTTPS enforced

15 pts

All traffic must be served over TLS encryption

Impressum present

15 pts

German-law legal notice page reachable from the homepage

Datenschutzerklärung present

15 pts

Privacy policy page available in German language

No pre-consent tracking

12 pts

No Google Analytics, GTM, Meta Pixel, Hotjar, or Microsoft Clarity loaded before user consent

No pre-consent external fonts

10 pts

No Google Fonts CDN or Adobe Typekit loaded before user consent

Cookie banner compliant

10 pts

Consent mechanism present with a visible reject/decline option

No mixed content

8 pts

No HTTP resources loaded on HTTPS pages (scripts, images, iframes)

EU-hosted infrastructure

7 pts

Application hosted on EU-region servers (Vercel EU, Hetzner, etc.)

Contact information present

5 pts

Email address or phone number findable on the site

Privacy policy localized

3 pts

German-language privacy content with DSGVO-specific terminology

Total: 100 points — weighted sum of all passed checks
08

Frequently asked questions about scoring

Common questions from builders and recruiters

How often are scores updated?

Quality scores update nightly via automated API checks. Traction data (visitors, uptime) updates every 6 hours. Efficiency scores update when verification data changes.

Can I game the score?

Every metric is pulled from external APIs — you can't self-report your way to the top. The only way to improve your score is to actually improve your project: better performance, proper security headers, DSGVO compliance, real users.

Why is DSGVO compliance weighted so heavily?

DSGVO compliance is table stakes for German B2B buyers. A technically excellent app that's non-compliant is unusable in the DACH market (Germany, Austria, Switzerland). This weighting reflects real-world hiring decisions.

Why use a logarithmic scale for visitor counts?

Linear visitor counts reward marketing budgets, not product quality. The log scale means going from 100 to 1,000 visitors is worth as much as going from 10,000 to 100,000. Building something people return to matters more than one viral post.

What if I disagree with my score?

File a dispute from your dashboard. We re-run the measurement, share the raw API responses, and explain the result. If we made an error, we correct it publicly.

Why isn't Efficiency available for Bronze tier projects?

Self-reported build stats are unreliable. The Efficiency leaderboard only includes projects where costs and token usage are verified via chat logs (Silver) or direct API integration with the AI tool (Gold).

Do you penalize specific AI coding tools?

No. Tool-specific leaderboards exist for comparison, but the scoring formulas are completely tool-agnostic. A Cursor project and a Lovable project are scored identically using the same metrics.

Which AI coding tools are supported?

We currently index projects built with Lovable, Bolt, v0, Cursor, Claude Code, and Replit. If a tool is missing, submit a request — we regularly add new ones.

How does this compare to Product Hunt or other directories?

Unlike Product Hunt, which relies on votes and editorial picks, every score here is computed automatically from external APIs. There is no editorial discretion, no subjective ranking. We measure technical quality, real usage, and compliance — not popularity contests.

What is the AI Code Smell detector?

Our custom scanner checks for 12 code patterns that AI coding tools commonly produce: excessive inline styles, missing React error boundaries, console.log statements left in production, hardcoded API keys, missing image alt text, oversized HTML documents, duplicate library loading, and more. Each smell is severity-weighted — critical issues like hardcoded secrets deduct 20 points, warnings deduct 8, and informational findings deduct 3.

How is Code Health scored for projects without public repos?

Code Health requires a public GitHub repository. Projects without one are scored using only the four core Quality metrics (Performance, Security, Compliance, Accessibility) at their original weights. This ensures projects aren't penalized for being closed-source.

Ready to see how your project scores?

We use Plausible Analytics — cookieless, DSGVO-compliant, EU-hosted. No tracking pixels, no fingerprinting. Privacy policy