Medo AppScore

Medo AppScore is an intelligent app analysis platform. It automatically evaluates any web application hosted on GitHub or GitLab.

Comment

Medo AppScore is an intelligent app analysis platform built for the Build with Medo Hackathon. It automatically evaluates any web application hosted on GitHub or GitLab, scores it across 7 quality dimensions, and generates concrete, prioritized improvement tips — all powered by Gemini 2.5 Flash.

Unlike static linters or manual checklists, AppScore uses AI to:

Score quality across 7 dimensions Generate actionable tips — concrete, step-by-step suggestions, not just scores Track progress across multiple re-analysis attempts Show you where you stand on the public leaderboard 7 Dimensions of Analysis Dimension Weight What It Measures Code Quality 15% Architecture, modularity, documentation Security 20% Vulnerabilities, secrets, dependency audits Efficiency 15% Performance, bundle size, async patterns Testing 15% Coverage, frameworks, CI integration Accessibility 10% WCAG compliance, ARIA, contrast Medo Integration 15% Auth, Data API, Analytics, Push Notifications User Experience 10% Onboarding, empty states, loading perception Key Features

  1. One-Click AI Analysis Paste your GitHub/GitLab repo URL, hit submit, and watch Gemini 2.5 Flash analyze your entire codebase. No config files, no manual setup.

  2. Improvement Tips (Not Just Scores) Every dimension comes with 2–3 actionable improvement tips. High-scoring apps still get tips to go from "good" to "great." Tips are sorted by impact and labeled with "Under 30 min" for quick wins.

  3. Re-Analyze After Changes Made improvements to your code? Click Re-analyze on your dashboard to trigger a fresh AI run on the same submission — no need to create a new attempt. Your score history is tracked automatically.

  4. Real-Time Dashboard Live score gauge, animated dimension bars, and real-time Supabase subscriptions keep your dashboard updated as soon as analysis completes.

  5. Public Leaderboard Compare your app's score with others across the platform. Filter by dimension (Overall, Security, Accessibility, Medo) and search by app name.

  6. Judge Review & Override Judges can review AI scores and manually override any dimension. A blend slider lets judges weight AI vs. manual scoring for the final result.

  7. Export Reports Export judge reviews as CSV or PDF for offline sharing and archival.

  8. File Uploads Upload pitch decks (PDF) and screenshots directly to Supabase Storage.

Architecture User → React App → Supabase Auth ↓ Supabase Postgres (11 tables, RLS) ↓ Supabase Edge Functions (Deno) ├── analyze-submission → Gemini 2.5 Flash └── compute-leaderboard → Aggregation queries ↓ Supabase Realtime (live score updates)

Database Schema submissions — App metadata, repo URL, status analysis_results — AI-generated dimension scores + overall grade findings — Actionable improvement tips per dimension leaderboard_entries — Ranked entries with top dimension tracking judge_reviews — Manual overrides with AI blend ratio notifications — Real-time user alerts via Supabase Realtime profiles — User roles (participant / judge / admin) teams — Auto-created teams per user

Dashboard — Score Overview An animated circular gauge shows your overall score and grade. Horizontal bars break down each of the 7 dimensions.

Improvement Tips Tab Tips are grouped by dimension. Each tip is a concrete, actionable suggestion — no abstract "points." Quick wins are highlighted for changes under 30 minutes.

Attempt History Tab A Recharts line chart shows your score trajectory across re-analysis attempts. A comparison table highlights per-dimension improvements.

Leaderboard Real-time table with filtering by dimension, search by app/team, and a "Latest only" toggle.

Judge Dashboard Judges see all submissions, review AI scores, adjust any dimension with a slider, and leave comments. A blend ratio lets them weight AI vs. manual scoring. uploaded_files — File metadata linking to Supabase Storage

Built With

Share this project:

Updates