🧬 EvolvixOS v9.2

The open-source, self-hosted AI engineering platform.

100% local core · zero paid tokens · 44 tools · 81 models · 35K APIs

🌐 Live Platform · 📚 Docs · 🎨 Studio · 📦 Models · 🔑 Developer Portal

---

✨ What is EvolvixOS?

EvolvixOS is a fully self-hosted AI engineering platform that runs on a single server. It features Mr James — an autonomous AI agent with 44 tools, triple-brain routing (Groq + Gemini + Kimi), and access to 35,277 searchable APIs. Zero paid tokens for core logic. Everything runs locally with optional external AI engines for speed and multimodal capabilities.

🧠 Triple-Brain Routing

Mr James automatically selects the best AI engine for each task:

Engine	Model	Speed	Context	Primary Use
🟢 Groq	gpt-oss-120b	467 tok/s	128K	Tool-use precision, agentic execution
🔵 Gemini	gemini-3.6-flash	Fast	1M	Vision, multimodal, TTS, large context
🟡 Kimi	moonshot-v1-32k	Medium	32K	Complex reasoning fallback
⚪ Ollama	qwen2.5:14b/7b/3b	CPU	32K	Local, offline, zero-cost fallback

🔧 44 Tools

Click to expand full tool list

Category	Count	Tools
📁 File Operations	6	file_read, file_write, file_edit, file_list, file_delete, code_analyze
💻 Code Execution	3	python_exec, bash_exec, sandbox_exec (CubeSandbox MicroVM)
🤖 AI / LLM	4	call_free_llm, gemini_vision, gemini_tts, file_upload
🌐 Smart API	3	api_auto_route (35K APIs), smart_api_call, http_request
☁️ Tencent Cloud	1	tencent_cloud (CVM, CDB, VPC, SSL, DNSPod, CDN, Billing, CAM, Hunyuan, AIArt)
💬 TIMSDK Chat	4	tim_send_message, tim_create_group, tim_send_group_message, tim_import_user
🧠 Team Memory	2	team_memory_search, team_memory_save (TencentDB)
🎭 Agent Library	2	search_subagents (217 templates), set_persona (16 MBTI types)
⚙️ System	3	get_system_info, manage_services, get_service_logs
🔧 Other	16	web_search, image_gen, and more

🏗️ Architecture

┌─────────────────────────────────────────────────────────────────┐
│                      EvolvixOS v9.2 Platform                     │
├──────────────────┬──────────────────┬───────────────────────────┤
│   Mr James v9.2   │    Model API     │      Auth API             │
│   44 tools        │    :5010         │      :5000                │
│   Triple-brain    │    81 models     │      JWT + OTP            │
│   routing         │    Groq+Gemini   │      SHA-256 API keys     │
│                   │    +Kimi+Ollama  │      Rate limiting        │
├──────────────────┼──────────────────┼───────────────────────────┤
│  Discovery Engine │  Dashboard       │   Developer Portal        │
│  Hourly GitHub    │  evolvixos.com   │   /developer              │
│  4 repos synced   │  /studio         │   API key management      │
│  35K APIs indexed │  /models         │   API documentation       │
├──────────────────┴──────────────────┴───────────────────────────┤
│                     Backend Services                             │
├───────────────┬──────────────┬──────────────┬───────────────────┤
│ TencentDB     │ CubeSandbox  │  TIMSDK      │  Octop            │
│ Agent Memory   │ MicroVM      │  Chat SDK    │  217 subagents    │
│ 3 containers   │ (Docker)     │  1K MAU     │  16 MBTI profiles │
├───────────────┴──────────────┴──────────────┴───────────────────┤
│                   Tencent Cloud SDK                              │
│   Go binary (tccli 9.3MB) + Python SDK (12 services)             │
│   CVM · CDB · VPC · SSL · DNSPod · CDN · Billing · CAM · Hunyuan │
├──────────────────────────────────────────────────────────────────┤
│              Infrastructure (16 systemd services)                 │
│  Nginx HTTPS · Let's Encrypt SSL · Ollama · Docker              │
│  Server: Hetzner Cloud (16 vCPU, 30GB RAM, 600GB disk)          │
└──────────────────────────────────────────────────────────────────┘

🔌 Integrations

☁️ Tencent Cloud SDK

• Go binary (tccli, 9.3MB): 7 services, 14 actions
• Python SDK: 12 services — comprehensive API coverage
• Services: CVM, CDB, VPC, SSL, DNSPod, CDN, Billing, CAM, Hunyuan, AIArt
• Config: TENCENTCLOUD_SECRET_ID + TENCENTCLOUD_SECRET_KEY

🤖 Octop — Self-Hosted AI Assistant

• 217 subagent templates across 16 categories (Engineering: 33, Specialized: 53, Marketing: 36, Security: 10)
• 16 MBTI personality profiles with behavior mappings
• 10 expert agent templates
• SSRF Guard (CWE-918 mitigation)

📦 CubeSandbox — MicroVM Execution

• Isolated code execution sandbox for AI agents
• Docker fallback mode (KVM not available on Hetzner Cloud)
• Pre-installed: numpy, pandas, scikit-learn, matplotlib
• Full MicroVM mode on CCX dedicated CPU servers

💬 TIMSDK — Real-Time Messaging

• Tencent IM SDK · 1,000 MAU free tier · REST API + Web UIKit
• Config: TIM_SDK_APP_ID + TIM_SECRET_KEY

🧠 TencentDB Agent Memory

• 3 Docker containers (core:8420, hub:8125, proxy:8096)
• Team memory with full-text search
• Powered by Groq for memory processing

✨ Google Gemini 3.6 Flash

• 37 models · 1M context · Vision + Multimodal + TTS
• gemini_vision: image analysis, OCR, chart reading
• gemini_tts: text-to-speech

⚡ Groq Integration

• gpt-oss-120b at 467 tok/s
• Primary execution engine for agentic tool-use
• Auto-fallback to Ollama when overloaded

🔍 Discovery Engine

Hourly GitHub scans across 4 repositories — 35,277 APIs indexed:

Repository	Content	Count
OpenClaw API Directory	APIs across 18 categories	25,822
API Mega List	Public API registries	7,000+
AI Agent Tools	Tools for AI agent development	84
Free LLM APIs	Models across 31 providers	442+

🌐 API Reference

Click to expand API endpoints

Endpoint	Method	Auth	Description
/api/health	GET	—	System health check
/api/agent/stream	POST	JWT/API	Streaming agent response
/api/agent	POST	JWT/API	Non-streaming agent response
/api/models	GET	—	List all registered models
/api/upload	POST	JWT	File upload (50MB max, multipart)
/api/docs	GET	—	API documentation
/auth/register	POST	—	User registration (OTP)
/auth/login	POST	—	User login (OTP)
/auth/api-keys/generate	POST	JWT	Generate API key
/auth/api-keys/list	GET	JWT	List API keys
/auth/api-keys/revoke	DELETE	JWT	Revoke API key
/auth/api-keys/usage	GET	JWT	API key usage stats

🔐 Security

Feature	Details
Authentication	JWT tokens with OTP verification
API Keys	SHA-256 hashed, evx_ prefix, per-user
Rate Limiting	30/min per IP, 100/min per API key
SSRF Guard	CWE-918 mitigation (ported from Octop)
Shell Injection	shlex.split, shell=False
Error Handling	21 blocks patched

📦 Frontend Pages

Page	URL	Description
Landing	/	Platform overview with architecture
Studio	/studio	Zero-knowledge dashboard
Models	/models	Visual model browser (81 models, 8 categories)
APIs	/apis	Searchable API directory (35K APIs)
Learn	/learn	Learning Hub (19-module full-stack course)
Developer	/developer	API key management & docs
Auth	/auth	Login/signup with OTP
Memory	memory.evolvixos.com	Team memory panel

🚀 Quick Start

# Clone
git clone https://github.com/Protremix/EvolvixOS.git
cd EvolvixOS

# Deploy (requires Ubuntu 22.04+)
chmod +x gpu_deploy.sh
./gpu_deploy.sh

# Or manual setup
pip3 install -r requirements.txt
python3 models/model_api.py &  # Model API on :5010
python3 auth/auth_api.py &      # Auth API on :5000
python3 dashboard/server.py &   # Dashboard on :8080

⚙️ Configuration

# Required for core
export JWT_SECRET="your-secret"

# Optional: External AI brains
export GROQ_API_KEY="your-groq-key"
export GEMINI_API_KEY="your-gemini-key"
export KIMI_API_KEY="your-kimi-key"

# Optional: Tencent Cloud
export TENCENTCLOUD_SECRET_ID="your-id"
export TENCENTCLOUD_SECRET_KEY="your-key"

# Optional: Tencent IM
export TIM_SDK_APP_ID="your-app-id"
export TIM_SECRET_KEY="your-secret"

📊 Platform Stats

Models	81 registered (Ollama + Discovery + Built-in)
APIs/Tools	35,277 searchable across 33 categories
Services	16 systemd services (all stable)
Server	Hetzner Cloud (16 vCPU, 30GB RAM, 600GB disk)
Stack	Python 3.14, Go 1.23.4, Nginx, Docker, systemd
SSL	Let's Encrypt (valid until Nov 2026)

🤝 Contributing

Contributions are welcome! Please read the contributing guidelines first.

📄 License

MIT — EvolvixOS is free and open-source software.

⭐ Star · 🍴 Fork · 📖 Docs · 🌐 Live