About CodeCortex Chat
CodeCortex Chat — a fast, private, streaming AI chat powered by OpenRouter. Markdown rendering, code highlighting, thinking mode, multi-language UI.
CodeCortex Chatis a privacy-friendly, streaming-first AI chat application built on top of the OpenRouter API. It is open-source, free to use, and designed to be lightweight enough to run on Vercel's free tier without compromising on the things that matter: speed, privacy, and a clean writing surface.
What makes it different
- Server-side API key. Your OpenRouter API key never reaches the browser. All model calls are relayed through a single server route.
- Local-first conversations. Your chat history lives in your browser's
localStorage, not in our database. Nothing leaves your device until you press Send. - No trackers. No Google Analytics, no Hotjar, no third-party cookies. The site honours Do Not Track.
- Multilingual. The UI auto-detects your language from your IP country and supports 10 locales including English, Japanese, Chinese, Korean, Spanish, French, German, Portuguese, Russian, and Arabic (RTL).
- Markdown-native. Tables, lists, fenced code blocks with syntax highlighting, copy buttons, and a Thinking mode that exposes the model's reasoning.
How it works
Browser ──POST /api/chat──► Next.js API route ──► OpenRouter API
◄──SSE stream────── (passthrough) ◄──SSE stream────The server route streams Server-Sent Events directly from OpenRouter to your browser, so the first token appears in about a second and the response flows in continuously. The conversation is trimmed to the last 8 turns before being sent, and the system prompt is pinned at the front of every request so the upstream provider's prompt cache can cut input cost where supported.
Tech stack
- Next.js 14 (App Router) + TypeScript
- Tailwind CSS for styling (single accent colour, neutral palette)
- react-markdown + remark-gfm + rehype-highlight for rendering
- React hooks for state (no Redux / Zustand until we need them)
- localStorage abstraction layer ready for a future DB migration
Privacy & safety
We do not maintain user accounts. The Service has no database of conversations. The only personal data we touch is your IP address, which we use for rate limiting and locale detection and discard immediately after the request completes. Markdown rendering is hardened against XSS: raw HTML,<script>,<iframe>, and similar elements are stripped before display. See our Privacy Policyfor full details.
Pricing
CodeCortex Chat is free to use. We may introduce advertising (e.g., Google AdSense) to cover infrastructure costs. We will never sell your data.
Contact & legal
Visit our Contact page for email addresses, or see the legal pages linked in the footer: Terms, Privacy, Cookies, Disclaimer, DMCA.