Not A Wrapper is an open-source, multi-AI chat platform with a unified interface for 80+ models across 8 providers. Bring your own API keys, connect MCP tools, upload files, and organize conversations into projects — all from a single, clean interface.
Forked from Zola and rebuilt with Convex, WorkOS AuthKit, and the Vercel AI SDK.
- Multi-provider AI chat — Stream responses from OpenAI, Anthropic, Google, Mistral, xAI, Perplexity, DeepSeek, and OpenRouter
- Streaming with reasoning — See model thinking in real-time (Claude, o3, DeepSeek R1, etc.)
- File uploads — Share documents, images, and code for AI analysis (Convex-backed storage)
- Guest access — Try the app without signing up (5 messages/day, limited model selection)
- Web search — Native provider search (OpenAI, Anthropic, Google, xAI) with Exa fallback for providers without built-in search
- MCP support — Connect external tool servers via the Model Context Protocol; per-tool approval, circuit breaker, and audit logging
- BYOK (Bring Your Own Key) — Securely use your own API keys with AES-256-GCM encryption at rest
- Projects — Group related chats into folders
- Public sharing — Publish chats with a shareable link and OG metadata
- Pinned chats — Pin important conversations for quick access
- Chat history — Full conversation history with search
- Light / Dark / System themes — Automatic or manual theme switching
- Layout options — Sidebar or fullscreen modes
- Favorite & hidden models — Curate your model list
- Custom system prompts — Set default instructions for AI behavior
| Provider | Notable Models | Capabilities |
|---|---|---|
| OpenAI | GPT-5.2, GPT-5.1, GPT-5, GPT-4.1, o3, o4-mini | Vision, Tools, Web Search |
| Anthropic | Claude Opus 4.6, Sonnet 4.5, Sonnet 4, Haiku 4.5 | Extended thinking, Vision, Web Search |
| Gemini 2.5 Pro, 2.5 Flash, 3 Pro Preview, Gemma 3 | Vision, Multimodal, Web Search | |
| Mistral | Mistral Large, Codestral, Pixtral Large, Ministral | Vision, Code |
| xAI | Grok 4, Grok 4.1 Fast, Grok Code Fast, Grok 2 Vision | Vision, Web Search |
| Perplexity | Sonar Pro, Sonar Deep Research, Sonar Reasoning Pro | Built-in web search |
| DeepSeek | DeepSeek R1, DeepSeek-V3 | Reasoning (via OpenRouter) |
| OpenRouter | 18+ models from all major providers | Aggregator, BYOK |
Models are regularly updated. The full list is configured in
lib/models/.
git clone https://github.com/darknightdesigner/not-a-wrapper.git
cd not-a-wrapper
bun install
cp .env.example .env.local # Edit with your keys
bun run env:check
bun dev # Starts Next.js + Convex dev serversThe app runs at http://localhost:3000. You need at least one AI provider key to chat.
For full setup (WorkOS AuthKit, Convex database, BYOK encryption), see INSTALL.md and docs/environment.md.
The current Convex setup uses the official WorkOS AuthKit component for webhook-backed app user sync on user.created, user.updated, and user.deleted.
| Layer | Technology |
|---|---|
| Framework | Next.js 16 (App Router), React 19, TypeScript |
| Database | Convex — Real-time reactive database with file storage and WorkOS webhook-backed user sync |
| Auth | WorkOS AuthKit — Hosted authentication with guest access preserved |
| AI | Vercel AI SDK v6 — Multi-provider streaming, tool calling |
| State | Zustand + TanStack Query |
| UI | Base UI + Tailwind CSS 4 |
| Observability | Sentry for app/error tracing, PostHog for product and LLM analytics, Braintrust for AI traces and evals |
| Testing | Vitest |
app/ # Next.js App Router
├── api/ # API routes (chat streaming, models, keys, MCP, projects)
├── c/[chatId]/ # Chat pages
├── p/[projectId]/ # Project pages
├── share/[chatId]/ # Public share pages
└── components/
├── chat/ # Chat UI, message rendering, tool invocations
├── layout/ # Sidebar, settings, dialogs
└── history/ # Chat history
lib/ # Shared logic
├── models/ # AI model definitions per provider
├── openproviders/ # Provider factory & SDK mapping
├── tools/ # Web search (provider + Exa), tool types
├── mcp/ # MCP server loading, circuit breaker
├── chat-store/ # Chat & message state (Zustand)
├── ai/ # Message conversion, context management
├── encryption.ts # AES-256-GCM for BYOK keys
└── config.ts # Rate limits, free models, defaults
components/ # Shared UI components (Base UI primitives)
convex/ # Database schema, queries, mutations, file storage
| Feature | Status |
|---|---|
| Multi-provider chat (8 providers, 80+ models) | Shipped |
| BYOK with AES-256-GCM encryption | Shipped |
| File uploads with Convex storage | Shipped |
| Projects & chat organization | Shipped |
| Public chat sharing | Shipped |
| Web search (native + Exa fallback) | Shipped |
| MCP tool integration | Shipped |
| Light/dark themes, layout options | Shipped |
| Sentry, PostHog, and Braintrust observability | Shipped |
| Guest access with rate limiting | Shipped |
| Code execution tools | Planned |
bun run dev # Dev server (Next.js + Convex)
bun run dev:clean # Dev with fresh .next cache
bun run env:check # Validate local environment variables
bun run lint # ESLint
bun run typecheck # TypeScript checks
bun run build # Production build (deploys Convex)
bun run test # VitestContributions are welcome. Please open an issue first to discuss what you'd like to change.
This project is a fork of Zola, the open-source AI chat interface. Special thanks to the Zola team for creating an excellent foundation.