Skip to content

About

AI-powered engineering analytics for GitHub — classify commits, PRs, and reviews with LLMs, score contributors across Impact, Quality, Collaboration & Consistency, and surface team health insights.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

GitRanked

AI-powered engineering analytics for GitHub repositories. Go beyond basic commit counts — GitRanked classifies every piece of work using AI, scores contributors across four dimensions, and surfaces actionable insights about your engineering team.

GitRanked repo analysis overview — facebook/react leaderboard, team health, contributor profiles, and work areas

Live demo →  ·  Ranked real repo (facebook/react)

Next.js TypeScript Tailwind CI License Deployed on Vercel


What it does

  • Evidence-Based Work Extraction — Commits and PRs are split into shipped capability units using file paths, patches, commit history, PR context, and LLM analysis. Each unit records its lifecycle role (foundation, advancement, refinement, or repair) and reconciles with previous passes.
  • Repo-Goal-Anchored Scoring — A per-repo goal tree is built from the README and merged PR titles; every capability node carries a deterministic centrality (1–5) for how core it is to the repo's purpose. A stateful capability ledger (UNIQUE per repo+capability) enforces cross-event dedup, and per-node saturating progress credit makes scores granularity-insensitive — the same work scores the same whether shipped as one big PR or five small ones, and foundational work outranks peripheral tweaks.
  • Multi-Dimensional Scoring — Contributors are scored across Impact, Quality, Collaboration, and Consistency with both current (decay-weighted) and all-time profiles, plus a 0–100 percentile so a great score means the same thing in any repo. Co-authored work is credited proportionally and PR commits are never double-counted.
  • Repository Health Metrics — Delivery, Collaboration, Code Quality, Review Health, and Knowledge Distribution on a 0–100 scale, computed from the same AI-classified work-unit model that powers contributor scores.
  • AI-Generated Summaries — Repository overviews, team insights, contributor profiles, impact analyses, and weekly/monthly reports
  • Shareable Dashboards — Token-based read-only sharing for stakeholders
  • Public Analytics Pages — SEO-optimized landing pages for GitHub insights, PR review metrics, repository health, and more

How it works

  1. Connect — Install the GitRanked GitHub App on your repos, or add public repos directly
  2. Ingest — Webhooks stream live events; historical backfill pulls the last 90 days
  3. Classify — pg-boss workers classify raw events into work units using AI
  4. Score — The scoring engine computes per-contributor scores with time-decay profiles
  5. Visualize — Interactive dashboards with leaderboards, health radar charts, collaboration networks, and AI-generated reports

Tech stack

Layer Technology
Frontend Next.js 16 (App Router), React 19, Tailwind CSS v4, Framer Motion, Recharts
Backend Next.js API Routes, NextAuth.js v5 (GitHub OAuth + App)
Database Neon PostgreSQL (serverless)
Queue pg-boss (background job processing)
AI OpenAI-compatible providers — defaults to OpenCode Go with DeepSeek V4 Flash when configured
Auth GitHub OAuth App + GitHub App (installation tokens)
Deployment Vercel (app) + GCP VM (worker)

Getting started

Prerequisites

  • Node.js 20+
  • A Neon PostgreSQL database
  • A GitHub OAuth App and GitHub App
  • An OpenCode Go or OpenRouter API key

Environment variables

# Database
DATABASE_URL=postgresql://...

# GitHub OAuth App
GITHUB_CLIENT_ID=...
GITHUB_CLIENT_SECRET=...
AUTH_GITHUB_ID=...      # same as GITHUB_CLIENT_ID
AUTH_GITHUB_SECRET=...  # same as GITHUB_CLIENT_SECRET

# GitHub App
GITHUB_APP_ID=...
GITHUB_PRIVATE_KEY=...  # PEM-encoded private key
GITHUB_WEBHOOK_SECRET=...

# AI provider (OpenCode Go is the default when this key is present)
OPENCODE_GO_API_KEY=...
OPENCODE_GO_MODEL=deepseek-v4-flash

# Optional OpenRouter fallback
OPENROUTER_API_KEY=sk-or-v1-...
OPENROUTER_MODEL=nvidia/nemotron-3-super-120b-a12b:free

# NextAuth
AUTH_SECRET=...         # generate with: openssl rand -hex 32
AUTH_URL=http://localhost:3000

# App
NEXT_PUBLIC_APP_URL=http://localhost:3000
NEXT_PUBLIC_GITHUB_APP_SLUG=your-app-slug

Development

npm install
npm run dev        # starts on http://localhost:3000
npm run worker     # starts pg-boss worker (in a separate terminal)

Database

npm run db:migrate  # runs schema migration

Testing

npm run test        # vitest
npm run typecheck   # tsc --noEmit
npm run lint        # eslint

Architecture

GitHub Events (webhooks / backfill)
        │
        ▼
   API Routes ────► Neon PostgreSQL
        │
        ▼
   pg-boss Queue ─► Worker (classification)
        │
        ▼
   OpenCode Go / OpenRouter AI ──► Evidence-backed Work Units + Scores
        │
        ▼
   Dashboard / APIs / Shared Links

License

MIT

About

AI-powered engineering analytics for GitHub — classify commits, PRs, and reviews with LLMs, score contributors across Impact, Quality, Collaboration & Consistency, and surface team health insights.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages