An AI-powered GitHub pull request reviewer that automatically analyzes code changes and posts actionable review comments on your PRs. Connect your GitHub App, and every opened, updated, or reopened pull request gets a structured markdown review — no manual sync required.

Chai AI Code Reviewer is a full-stack SaaS app built with Next.js. It listens for GitHub pull request webhooks, fetches the PR diff, uses vector search to find the most relevant code chunks, and generates a review with the Vercel AI SDK and OpenRouter. Reviews are posted back to GitHub as PR comments.

Optionally, you can sync an entire repository into Pinecone for richer context — for example, how changed code relates to other files outside the diff. PR reviews work without syncing; repo sync is an enhancement for deeper, cross-file feedback.

The dashboard lets you connect GitHub, browse repositories, track review history, monitor usage, and manage a Free or Pro subscription.

• Automatic PR reviews — Reviews run on opened, synchronize, and reopened PR events via GitHub webhooks
• AI-generated feedback — Structured markdown reviews covering correctness, security, performance, reliability, readability, and maintainability
• GitHub App integration — Install the app on your org or account to access repositories and post review comments
• Vector-based context — PR diffs are chunked and indexed in Pinecone so the model focuses on the most relevant changes
• Optional repo sync — Manually sync a repository’s codebase for extra context from files outside the PR diff
• Dashboard — Overview stats, repository list, pull request history, and GitHub connection status
• Review activity tracking — See recent reviews and their status (pending, processing, reviewed, rate-limited)
• Authentication — Sign in with GitHub using Better Auth
• Subscription billing — Free plan (5 reviews/month, public repos) and Pro plan (unlimited reviews, private repos) via Razorpay
• Background jobs — Long-running sync and review tasks handled reliably with Inngest
• Dark mode — Theme support across the dashboard UI

• Node.js 20+
• PostgreSQL database
• GitHub OAuth App + GitHub App
• Pinecone index (integrated embeddings: llama-text-embed-v2, field map text=text)
• OpenRouter API key
• Razorpay account (for subscriptions)

Copy .env.example to .env and fill in your credentials:

cp .env.example .env

npm install
npx prisma migrate dev
npm run dev

For background jobs locally, run the Inngest dev server in a separate terminal:

npx inngest-cli@latest dev

Set INNGEST_DEV=1 in your .env when using the local Inngest dev server.

Open http://localhost:3000, sign in with GitHub, connect the GitHub App, and open a pull request on a connected repository to trigger your first AI review.

app/                    # Next.js routes and API endpoints
features/
  billing/              # Razorpay subscriptions and usage limits
  dashboard/            # Dashboard shell and navigation
  github/               # GitHub App, webhooks, installations
  overview/             # Dashboard overview and activity
  pull-requests/        # PR list and review display
  repo-sync/            # Optional full-repo indexing
  reviews/              # PR review pipeline (Inngest + AI)
components/             # Shared UI components
lib/                    # Auth, DB, server actions
prisma/                 # Schema and migrations