CodeSentinel-AI

Code Review Bot is a modular monorepo that automates code reviews using AI, provides a developer UX for submitting and browsing reviews, and includes integrations for running checks locally or in CI. This README explains the architecture, how to run the system locally, a short demo script for a hackathon pitch, and deployment recommendations to help you win the hackathon.

Repository layout

• artifacts/codesentinel: UI frontend (Vite + React). Primary demo surface.
• api-server: Backend API and AI orchestration (Express/Node/TS). Hosts endpoints used by the frontend and any integrations.
• lib/api-client-react: Shared TypeScript API client used by the frontend.
• db: Drizzle schema and DB utilities for persisting conversations, findings, and review metadata.
• integrations / integrations-openai-ai-server: connectors for AI providers and batch/image/audio helpers.
• scripts: helper scripts and small utilities used during development.

High-level architecture

• Frontend (artifacts/codesentinel) — React app that collects code or PR links, sends them to the API, and displays structured findings and suggested fixes.
• API (api-server) — Receives requests, prepares prompts, orchestrates calls to AI providers, persists results in db, and returns structured findings.
• Shared libs (lib/*) — Reusable clients, types, and utilities used across frontend and backend.
• Integrations — Provider adapters and optional connectors for GitHub/GitLab/CI.

Why this wins

• End-to-end: UI + backend + persistence + provider adapters are present, so judges can see a complete flow.
• Extensible: adapter architecture makes it easy to swap AI providers or add a CI integration during the hackathon.
• Developer-friendly demo: local dev mode runs fast, and artifacts/codesentinel is crafted to be a clear, visual demo surface.

Prerequisites

• Node.js 18+ and pnpm (pnpm is required for workspace handling)
• Git (to clone)

Quickstart (recommended — full workspace)

1. Clone the repo and install dependencies at the workspace root:

git clone <repo-url>
cd Code-Review-Bot
pnpm install

2. Run the API and frontend from the workspace root (parallel terminals):

# Start the API
pnpm --filter @workspace/api-server run dev

# Start the frontend (codesentinel)
pnpm --filter @workspace/codesentinel run dev

Notes: artifacts/codesentinel references local workspace packages (e.g. @workspace/api-client-react) so running from the workspace root with pnpm install ensures the workspace links are resolved automatically.

Quickstart (frontend-only, best-effort)

• If you want to iterate on UI only: open artifacts/codesentinel, run pnpm install there, then pnpm run dev. Some workspace-linked packages may need to be built or copied if the root workspace isn't available.

Configuration

• Environment variables are read by the API and integrations. Create a .env file in api-server with provider keys and DB settings. Example keys:

OPENAI_API_KEY=...
DATABASE_URL=sqlite:./db/dev.sqlite

Data model and persistence

• db contains Drizzle schema for conversations, messages and findings. The API persists AI responses and review metadata so the UI can present history and diffs.