Topsha — AI skill for Claude Code

🐧 Topsha Local AI Agent for simple every day tasks

**AI Agent Framework for Self-Hosted LLMs — deploy on your infrastructure, keep data private.**

🎯 **Built for companies and developers who need:**

• 100% on-premise AI agents (no data leaves your network)
• Any OpenAI-compatible LLM (vLLM, Ollama, llama.cpp, text-generation-webui)
• Production-ready security (battle-tested by 1500+ hackers)
• Simple deployment (docker compose up and you're done)

---

Why LocalTopSH?

🏠 100% Self-Hosted

Unlike cloud-dependent solutions, LocalTopSH runs entirely on your infrastructure:

Problem	Cloud Solutions	LocalTopSH
Data Privacy	Data sent to external APIs	✅ Everything stays on-premise
Compliance	Hard to audit	✅ Full control, easy audit
API Access	Need OpenAI/Anthropic account	✅ Any OpenAI-compatible endpoint
Sanctions/Restrictions	Blocked in some regions	✅ Works anywhere
Cost at Scale	$0.01-0.03 per 1K tokens	✅ Only electricity costs

🤖 Supported LLM Backends

Backend	Example Models	Setup
vLLM	gpt-oss-120b, Qwen-72B, Llama-3-70B	vllm serve model --api-key dummy
Ollama	Llama 3, Mistral, Qwen, 100+ models	ollama serve
llama.cpp	Any GGUF model	llama-server -m model.gguf
text-generation-webui	Any HuggingFace model	Enable OpenAI API extension
LocalAI	Multiple backends	Docker compose included
LM Studio	Desktop-friendly	Built-in server mode

💰 Cost Comparison (1M tokens/day)

Solution	Daily Cost	Monthly Cost
OpenAI GPT-4	~$30	~$900
Anthropic Claude	~$15	~$450
Self-hosted (LocalTopSH)	Electricity only	~$50-100 (GPU power)

🌍 Works Everywhere

• ✅ Russia, Belarus, Iran — sanctions don't apply to self-hosted
• ✅ China — no Great Firew