helwod/deepwhale repository overview

⁠Source Repository

https://github.com/bchhngsaygez/deepwhale⁠

⁠DeepWhale — DeepSeek Web-to-API Bridge

DeepWhale converts DeepSeek Chat⁠'s web interface into a fully OpenAI-compatible API endpoint with a built-in admin dashboard for real-time management, testing, and monitoring.

It uses browser fingerprinting via cloakbrowser (Playwright) to bypass Cloudflare and bot-detection, enabling reliable programmatic access to DeepSeek's models.

NOTE: I have no idea why this repo of mine runs perfectly—smooth as butter—on Cline, Roo Code, Continue.dev, and similar tools, yet when I switch to CLI AI agents, it absolutely refuses to work. It can't do a damn thing—creating files, writing code, making folders—nothing works. If you know how to fix this, please feel free to leave suggestions or a solution in the "Issues" section.

⁠Features

• OpenAI-Compatible API — Drop-in replacement for https://api.openai.com/v1. Works with any OpenAI client (Cline, Roo Code, Continue.dev, OpenRouter, etc.).
• Admin Dashboard — Modern web UI with language switcher (English / Vietnamese), real-time logs, multi-turn chat, interactive API tester, and live server/account management.
• Cloudflare Bypass — Uses cloakbrowser with real Chromium fingerprinting. No manual cookie extraction or CAPTCHA solving.
• Proof-of-Work Solver — Automatically solves DeepSeek's PoW challenges using Web Workers in the browser, with a pure-Python fallback.
• Account Rotation — Configure multiple DeepSeek accounts; the server rotates them in round-robin, skipping failed logins automatically.
• Auto-Recovery — Detects expired or invalid tokens and re-authenticates transparently mid-request.
• Streaming & Non-Streaming — Full SSE streaming with text/event-stream and standard JSON responses. Supports auto-continue for long responses (up to 8 rounds).
• Thinking/Reasoning Support — Models like deepseek-reasoner emit thinking content surfaced in responses.
• Model Aliasing — Map any model name (gpt-4o, o1, qwen-plus, etc.) to a DeepSeek model.
• Multi-Turn Chat — Built-in chat UI with model selection and thinking mode toggle.
• Real-Time Logs — Live SSE-streamed logs with color-coded categories and auto-scroll.
• CORS Enabled — Access the API from browser-based tools and web UIs.

⁠Quick Start (Docker)

# Clone and enter the repository
git clone https://github.com/bchhngsaygez/deepwhale.git
cd deepwhale

# Create your .env file
cp .env.example .env
# Edit .env with your DeepSeek email/password

# Start with Docker Compose
docker compose up -d

Open http://localhost:5001/admin/login (default password: admin).

⁠Manual Setup

⁠Prerequisites

• Python 3.11+
• pip
• ~2 GB free disk space (for Chromium)

⁠Installation

pip install -r requirements.txt

Download the cloakbrowser Chromium build for your platform from the CloakBrowser releases page⁠.

⁠Configuration

# .env
[email protected]
DEEPSEEK_PASSWORD=your_password
API_KEY=sk-my-secret-key-1
PORT=5001
HOST=0.0.0.0
ADMIN_PASSWORD=admin
FLASK_SECRET=your-random-secret-here

⁠Start the Server

python server.py

⁠Test via Command Line

curl http://localhost:5001/v1/models \
  -H "Authorization: Bearer sk-my-secret-key-1"

curl http://localhost:5001/v1/chat/completions \
  -H "Authorization: Bearer sk-my-secret-key-1" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

⁠API Endpoints

⁠Admin Dashboard

The dashboard features five sections accessible from the sidebar:

Language switching between English and Vietnamese is available in the top-right corner of the dashboard.

⁠Supported Models

⁠Project Structure

deepwhale/
├── server.py              # Flask API + admin dashboard
├── deepseek_client.py     # Browser session, login, API calls
├── pow_solver.py          # PoW challenge solver
├── utils.py               # Shared utilities (UTF-8, env loading)
├── test_client.py         # End-to-end test script
├── templates/
│   ├── login.html         # Admin login (dark theme, i18n)
│   └── dashboard.html     # Admin dashboard (i18n EN/VI)
├── Dockerfile             # Containerized deployment
├── docker-compose.yml     # Docker Compose setup
├── requirements.txt       # Python dependencies
├── .env.example           # Environment template
└── README.md              # This file

⁠Star History

⁠

⁠License

MIT⁠