AI Code Reviewer

Real-time AI-powered code analysis running on Cloudflare's edge. Paste code, get a streaming review — security audits, performance analysis, documentation feedback, or a general quality pass.

Live: ai-code-reviewer-5fq.pages.dev

---

How It Works

Browser (React) ──WebSocket──> Cloudflare Worker ──> Durable Object ──> Workers AI (Llama 3.3 70B)
                                                            |
                                                     Persistent state
                                                     (review history)

• The frontend connects via WebSocket on load and streams review output chunk-by-chunk.
• The backend is a Cloudflare Durable Object — one persistent instance per agent that holds conversation history and past reviews in storage.
• Language detection runs before the AI call. If the selected language doesn't match the code, the review is rejected with a suggestion.

---

Stack

Layer	Technology
Frontend	React + TypeScript (Vite)
Backend	Cloudflare Workers (TypeScript)
State	Cloudflare Durable Objects
AI	Workers AI — Llama 3.3 70B Instruct
Deploy	Cloudflare Pages + Workers

---

Review Types

Type	What it checks
quick	Overall code quality, structure, naming, logic
security	Vulnerabilities, OWASP Top 10, injection risks
performance	Bottlenecks, algorithmic complexity, memory usage
documentation	Inline comments, docstrings, API surface clarity

---

Local Development

Prerequisites: Node.js 18+, a Cloudflare account, Wrangler CLI

# 1. Install dependencies
npm install
cd frontend && npm install && cd ..

# 2. Configure environment
cp .env.example .env
# Fill in CLOUDFLARE_ACCOUNT_ID and CLOUDFLARE_API_TOKEN

# 3. Start the backend Worker (terminal 1)
npm run dev

# 4. Start the frontend (terminal 2)
npm run dev:frontend

Frontend: http://localhost:5173 Worker: http://localhost:8787

---

WebSocket Protocol

Connect to /agent and send JSON messages:

Submit code for review:

{
  "type": "submit_code",
  "code": "func main() { fmt.Println(\"hello\") }",
  "language": "go",
  "category": "performance"
}

Server events:

{ "type": "stream", "text": "chunk..." }   -- streaming content
{ "type": "done",   "review": { ... } }    -- review complete
{ "type": "language_error", "error": "..." } -- language mismatch
{ "type": "error",  "error": "..." }       -- general error

List past reviews:

{ "type": "list_reviews" }

---

REST API

# Submit for review (non-streaming)
curl -X POST https://ai-code-reviewer-backend.mohammedfirdousaraoye.workers.dev/api/review \
  -H "Content-Type: application/json" \
  -d '{"code": "...", "category": "quick", "language": "python"}'

# Get all past reviews
curl https://ai-code-reviewer-backend.mohammedfirdousaraoye.workers.dev/api/reviews

# Health check
curl https://ai-code-reviewer-backend.mohammedfirdousaraoye.workers.dev/health

---

Deployment

# Deploy the Worker
npm run deploy

# Deploy the frontend to Cloudflare Pages
cd frontend && npm run deploy

# Verify after deploy
npm run verify

# Tail live logs
npm run logs

---

Keyboard Shortcuts

Shortcut	Action
Ctrl/Cmd + K	Focus code input
Ctrl/Cmd + Enter	Submit review

---

Environment Variables

# .env (never commit this)
CLOUDFLARE_ACCOUNT_ID=your-account-id
CLOUDFLARE_API_TOKEN=your-api-token

---

References

• Workers AI Models
• Durable Objects
• Cloudflare Pages