# Draftify

> Draftify is an AI-powered coding workspace that delivers Opus-class and Sonnet-class intelligence with no token limits, no message caps, and no cooldown periods. It provides unlimited AI-assisted development through a web dashboard, CLI, and a fully OpenAI-compatible Developer API — all backed by a zero-retention privacy policy.

## About

Draftify is built for developers who need deep, uninterrupted AI assistance for complex coding tasks. Unlike consumer AI chatbots that impose per-conversation token limits, 4-hour cooldowns, or daily message caps, Draftify provides a continuous, production-grade AI environment where context windows are massive and sessions never expire.

The platform runs on a proprietary hybrid architecture using coding-optimized AI models that benchmark at the same tier as the latest Opus and Sonnet class models. All prompt data is processed in-memory with a strict zero-retention policy — source code is never stored or used for model training.

- Website: <https://draftify.site>
- Documentation: <https://draftify.site/features>
- Terms of Service: <https://draftify.site/tos>
- Privacy Policy: <https://draftify.site/privacy>

## Core Features

- **Huge Context Window**: Feed massive codebases and documentation directly into the model without truncation. No artificial token limits.
- **Real-time Streaming (SSE)**: Zero-latency generation with token-by-token streaming via Server-Sent Events directly to your workspace.
- **Developer API Access**: OpenAI-compatible REST API at `https://draftify.site/api/v1/chat/completions` with Bearer token authentication. Integrates with Cursor, Claude Code, Kilo Code, Codex CLI, Trae, Grok CLI, and any OpenAI-compatible client.
- **Smart Artifact Generation**: Generate entire code files, React components, and HTML documents in a dedicated side-by-side preview window with live rendering.
- **Dynamic AI Skills System**: Modular knowledge base with on-demand skill loading. The AI only loads domain-specific instructions (frontend design, brand guidelines, theme generation, copywriting) when needed, keeping simple interactions fast and token-efficient.
- **Enterprise-grade Security**: Strict zero-retention policy. Prompts and source code are processed in-memory only and never stored, logged, or used for training.
- **Model-Agnostic Architecture**: Seamlessly switch between multiple premium model tiers optimized for different tasks.
- **CLI Integration**: Native command-line interface with device authentication, session history, and full streaming support.

## Available AI Models

Draftify offers multiple intelligence tiers, each benchmarked against frontier-class models:

| Model | Description | API Input Price | API Output Price |
|---|---|---|---|
| Opus 4.8-level | Highest intelligence tier. Complex reasoning, architecture, deep analysis. | €2.00 / 1M tokens | €5.00 / 1M tokens |
| Opus 4.7-level | Advanced reasoning and logic-heavy tasks. | €1.00 / 1M tokens | €3.00 / 1M tokens |
| Opus 4.6-level | Strong general-purpose coding intelligence. | €1.00 / 1M tokens | €2.00 / 1M tokens |
| Fable 5-level | Creative and narrative-focused generation. | €2.50 / 1M tokens | €6.00 / 1M tokens |
| Sonnet 5-level | Fast, efficient model for routine coding tasks. | €0.20 / 1M tokens | €0.74 / 1M tokens |

## Subscription Plans

All plans include unlimited tokens and no message caps.

| Plan | Monthly Price | Yearly Price | Includes |
|---|---|---|---|
| Starter | €40/mo | €400/yr | Unlimited Sonnet 5-level usage. Standard support. |
| Pro | €90/mo | €900/yr | Unlimited Sonnet 5-level + Opus 4.7-level. Priority support. |
| Scale | €130/mo | €1,300/yr | Unlimited Sonnet 5-level + Opus 4.7-level + Opus 4.8-level (chatbot) + Fable 5-level. 24/7 priority support. |

The Developer API is available to all users without a subscription on a pay-as-you-go basis. No minimum spend.

Yearly plans save approximately 16% compared to monthly billing.

## Developer API

Draftify exposes a fully OpenAI-compatible REST API for programmatic access:

- **Base URL**: `https://draftify.site/api/v1`
- **Chat Completions**: `POST /api/v1/chat/completions`
- **List Models**: `GET /api/v1/models`
- **Authentication**: Bearer token via `Authorization: Bearer <api-key>` header, or `x-api-key` header for Anthropic-style clients.
- **Streaming**: Full SSE streaming support via `"stream": true`.
- **Tool Use / Function Calling**: Native support for tool definitions and function calling.
- **CORS**: Enabled with `Access-Control-Allow-Origin: *` for browser-based integrations.

### IDE and Tool Integrations

The Draftify API is compatible with:

- **Cursor** — Use as a custom OpenAI-compatible provider
- **Claude Code** — Compatible via x-api-key authentication
- **Kilo Code** — Direct OpenAI-compatible integration
- **Codex CLI** — Set as custom base URL
- **Trae** — OpenAI-compatible provider
- **Grok CLI** — Custom endpoint support
- **Droid** — Direct integration
- **Any OpenAI-compatible client** — Standard REST API

## AI Skills System

Draftify uses a dynamic skills architecture where the AI loads specialized knowledge on demand:

- **General** — Default personality and response guidelines (always loaded)
- **Brand Guidelines** — Visual identity, tone of voice, and design system rules
- **Doc Co-authoring** — Collaborative document creation and editing patterns
- **Frontend Design** — Anti-slop design system with premium typography, color calibration, and layout rules
- **Theme Factory** — Preset styling themes (Arctic Frost, Botanical Garden, Desert Rose, Forest Canopy, Golden Hour, Midnight Galaxy, Modern Minimalist, Ocean Depths, Sunset Boulevard)
- **Web Artifacts Builder** — Self-contained React + Vite + Tailwind project scaffolding with bundled shadcn/ui components

## Security and Privacy

- **Zero-Retention Policy**: All prompts and code are processed in-memory only. Nothing is stored after generation completes.
- **No Training Data Usage**: User code and prompts are never used for model training or improvement.
- **Hashed API Keys**: API keys are stored as SHA-256 hashes. Raw keys are never persisted server-side.
- **Secure Payments**: All transactions processed by Whop with PCI-compliant infrastructure.

## Technical Stack

- **Framework**: Next.js 16 with Turbopack
- **Authentication**: NextAuth.js with credential-based and session-based auth
- **Database**: PostgreSQL via Prisma ORM
- **Payments**: Whop checkout integration with Stripe processing
- **Hosting**: Vercel Edge Network
- **AI Infrastructure**: Proprietary hybrid routing system with multiple frontier-class model backends

## Contact

- Website: <https://draftify.site>
- Support: Available through the platform dashboard