The Super AI Agent built from ClaudeCode + Hermes + OpenClaw. Local Brain, Voice-First, Three-Tier Memory. 80% of tasks complete without calling LLM APIs.
Complete feature comparison with OpenClaw and Hermes
| Feature | OpenClaw | Hermes | ZeusHammer |
|---|---|---|---|
| Local Brain | No | No | Yes |
| Voice-First Design | Partial | Partial | Yes |
| Three-Tier Memory | No | No | Yes |
| Meditation Mode | No | No | Yes |
| Deep Reflection | No | No | Yes |
| China LLM Native | Config | Partial | Yes |
| Tool Detector | No | No | Yes |
| OSV Malware Scan | No | No | Yes |
| Cost Tracking | No | Partial | Yes |
| WebRTC Audio | No | No | Yes |
| Skill Auto-Generation | No | No | Yes |
| Built-in Tools | 20+ | 40+ | 50+ |
| Homebrew Install | npm | pip | brew |
| Message Channels | 30+ | 20+ | 13+ |
| AI Models | 35+ | 15+ | 20+ |
| MCP Support | Yes | Yes | Yes |
| Browser Automation | Yes | Yes | Yes |
80% of tasks complete via intent recognition + skill matching locally, without calling LLM APIs. Significantly reduces cost and improves response speed.
Complete voice-first experience: wake word, Whisper STT, Edge TTS, auto language detection. Supports background daemon mode.
Short-term (LRU) + Long-term (SQLite FTS5) + Working memory, plus Chroma vector store. Most complete memory capabilities.
Built-in unified clients for DeepSeek, Qwen, GLM, Moonshot, ERNIE, Doubao, MiniMax via chinawhapi.com.
Quick start: curl -sSL https://raw.githubusercontent.com/pengrambo3-tech/ZeusHammer/master/install.sh | bash
Ecosystem scale (400+ contributors vs new project). Message channels (13+) less than OpenClaw (30+) and Hermes (20+). Community maturity needs verification.
Built from the best features of ClaudeCode, Hermes, and OpenClaw
Intent recognition and pattern matching. Auto-learns new skills. 80% of common tasks complete without LLM calls.
Self-LearningLocal Whisper STT + Edge TTS. True hands-free operation. No data sent to cloud. Fully offline capable.
Privacy FirstShort-term (LRU) + Long-term (SQLite) + Working memory. Persistent cross-session context. Never forget.
24/7 PersistentUnified API via chinawhapi.com for DeepSeek, Qwen, GLM, Moonshot, ERNIE, Doubao, MiniMax.
10+ ModelsChoose the installation method that works for you
curl -sSL https://raw.githubusercontent.com/pengrambo3-tech/ZeusHammer/master/install.sh | bashgit clone https://github.com/pengrambo3-tech/ZeusHammer.git
cd ZeusHammer
pip install -r requirements.txtdocker pull zeushammer/zeushammer
docker run -it zeushammer/zeushammergit clone https://github.com/pengrambo3-tech/ZeusHammer.git
cd ZeusHammer
pipx install .pip install git+https://github.com/pengrambo3-tech/ZeusHammer.git# Add to ~/.zeushammer/.env
OPENAI_API_KEY=your_key
API_BASE=https://api.chinawhapi.com/v1
MODEL=deepseek-chat# Homebrew
brew uninstall zeushammer
# pipx
pipx uninstall zeushammer
# Remove config
rm -rf ~/.zeushammerMultiple modes for different use cases
# Interactive mode
zeushammer --mode cli
# or
python -m src.main --mode cli
# Single query
zeushammer "Hello, write Python code"# Start web interface
zeushammer --mode web
# or
python -m src.main --mode web
# Access http://localhost:7860# Start voice interaction
zeushammer --mode voice
# or
python -m src.main --mode voice
# Say "Zeus" or "宙斯" to wake# Rich terminal interface
zeushammer --mode tui
# or
python -m src.main --mode tui# Safe mode (recommended)
zeushammer --level safe
# Semi-open
zeushammer --level semi_open
# Full open (beast mode)
zeushammer --level full_open# Edit config
nano ~/.zeushammer/config.yaml
# Available models:
# - deepseek-chat (recommended)
# - qwen-turbo
# - glm-4
# - claude-3-5-sonnet
# - gpt-4oAccess global mainstream LLMs via chinawhapi.com
V3, Coder
Turbo, Plus, Max
GLM-4, GLM-4V
8K, 32K, 128K
Bot 4.0, Long
Pro, Lite
ABAB6 Chat
Sonnet, Opus, Haiku
4o, Turbo
Local Models