All systems operational · 16/16 services running

The open-source AI
engineering platform

Self-hosted. Zero paid tokens for core logic. 44 tools, 81 models, 35,277 searchable APIs. Triple-brain routing with automatic fallback.

⭐ Star on GitHub 🌐 Visit evolvixos.com
44
Tools
81
Models
35K
APIs Indexed
3
AI Brains

Architecture

Everything runs on a single server. Fully self-contained.

┌─────────────────────────────────────────────────────────────────┐ │ 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) + 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) │ └──────────────────────────────────────────────────────────────────┘

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
Google 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

File ops, code execution, AI, APIs, cloud management, messaging, memory, and more

file_read
Read file contents
file_write
Write/create files
file_edit
Edit existing files
file_list
List directory contents
file_delete
Delete files
code_analyze
Analyze code with AI
python_exec
Execute Python code
bash_exec
Execute shell commands
sandbox_exec
CubeSandbox MicroVM execution
call_free_llm
Delegate to 442+ free LLM APIs
gemini_vision
Image analysis, OCR, charts
gemini_tts
Text-to-speech via Gemini
file_upload
Upload with Gemini Vision analysis
api_auto_route
Semantic API discovery (35K APIs)
smart_api_call
HTTP API execution
http_request
Generic HTTP requests
tencent_cloud
12 services: CVM, CDB, VPC, SSL, DNS...
tim_send_message
TIMSDK direct messaging
tim_create_group
TIMSDK group creation
tim_send_group_message
TIMSDK group messaging
tim_import_user
TIMSDK user import
team_memory_search
TencentDB full-text search
team_memory_save
TencentDB memory persistence
search_subagents
Search 217 agent templates
set_persona
Switch 16 MBTI personalities
get_system_info
Server status & specs
manage_services
Start/stop systemd services
get_service_logs
View service logs
web_search
Search the web
image_gen
AI image generation

Integrations (v9.2)

7 major integrations powering the platform

☁️
Tencent Cloud SDK
✅ Live · Go binary + Python
12 Python services + 7 Go binary services (9.3MB tccli binary for high-performance API calls). Go binary is tried first with automatic Python SDK fallback.
12Python services
7Go services
258Products
14Actions
CVM (servers) · CDB (MySQL) · VPC (networking) · SSL (certificates) · DNSPod (DNS) · CDN · Billing · CAM (users) · Hunyuan (LLM) · AIArt (image gen)
🤖
Octop — Self-Hosted AI Assistant
✅ Live · 217 subagents
217 subagent templates across 16 categories. 16 MBTI personality profiles with behavior mappings for answer style, casual chat, conflict, creativity, emotion, and planning.
217Subagent templates
16MBTI profiles
10Expert agents
16Categories
📦
CubeSandbox — MicroVM Execution
✅ Live · Docker mode
Isolated code execution sandbox for AI agents. Docker fallback mode (KVM not available on Hetzner Cloud). Full MicroVM mode on CCX dedicated CPU servers.
v0.6.0SDK version
4Pre-installed libs
DockerCurrent mode
💬
TIMSDK — Real-Time Messaging
⚡ Code ready · Needs credentials
Tencent IM SDK for real-time chat integration. 1,000 MAU free tier. Server-side REST API + Web UIKit. 4 tools: send message, create group, send group message, import user.
4Tools
1KMAU free tier
REST+ Web UIKit
🧠
TencentDB Agent Memory
✅ Live · 3 containers
Team memory with full-text search. 3 Docker containers: memory-core (:8420), memory-hub (:8125), proxy (:8096). Powered by Groq for memory processing.
3Docker containers
2Tools
GroqProcessing engine
✨
Google Gemini 3.6 Flash
✅ Live · 37 models
Vision, multimodal, and TTS capabilities. 1M token context window. gemini_vision for image analysis/OCR/charts. gemini_tts for text-to-speech. File upload analysis for images, PDFs, and documents.
37Models
1MContext window
2Tools
⚡
Groq — High-Speed Inference
✅ Live · 467 tok/s
gpt-oss-120b at 467 tokens/second. Primary execution engine for agentic tool-use. Auto-fallback to Ollama when overloaded.
467Tokens/sec
120BModel size
128KContext

Discovery Engine

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

📚
OpenClaw API Directory
25,822 APIs across 18 categories
Synced hourly
📋
API Mega List
7,000+ APIs from public registries
Synced hourly
🔧
AI Agent Tools
84 tools for AI agent development
Synced hourly
🆓
Free LLM APIs
442+ models across 31 providers
Synced hourly

API Reference

REST API for connecting EvolvixOS to external projects

EndpointMethodAuthDescription
/api/healthGET—System health check
/api/agent/streamPOSTJWT/APIStreaming agent response
/api/agentPOSTJWT/APINon-streaming agent response
/api/modelsGET—List all registered models
/api/uploadPOSTJWTFile upload (50MB max, multipart)
/api/docsGET—API documentation
/auth/registerPOST—User registration (OTP)
/auth/loginPOST—User login (OTP)
/auth/api-keys/generatePOSTJWTGenerate API key
/auth/api-keys/listGETJWTList API keys
/auth/api-keys/revokeDELJWTRevoke API key
/auth/api-keys/usageGETJWTAPI key usage stats

Security

Multi-layer protection for self-hosted AI

🔐
JWT + OTP
Token-based auth with OTP verification
🔑
API Keys
SHA-256 hashed, evx_ prefix
🛡️
Rate Limiting
30/min IP, 100/min API key
🚫
SSRF Guard
CWE-918 mitigation ported from Octop

Quick Start

Deploy EvolvixOS in minutes

# Clone the repository 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 # Configure (optional) export GROQ_API_KEY="your-groq-key" export GEMINI_API_KEY="your-gemini-key" export KIMI_API_KEY="your-kimi-key" export TENCENTCLOUD_SECRET_ID="your-tc-id" export TENCENTCLOUD_SECRET_KEY="your-tc-key"