Choose your deployment environment. Each scenario includes step-by-step instructions.
Kali Linux provides the best environment for RedClaw AI. The setup script installs all dependencies, creates a systemd service, and configures the agent for full system access.
# Clone from GitHub git clone https://github.com/musidd/redclaw.git cd redclaw
The setup script installs Python dependencies, configures the agent, creates a systemd service, and optionally installs common offensive tools.
sudo bash scripts/setup-kali.sh
This installs: Python 3.11, pip packages, Playwright, Go (for tool installs), and creates /opt/redclaw
nano /opt/redclaw/.env # Required: at least one LLM provider key OPENAI_API_KEY=sk-... ANTHROPIC_API_KEY=sk-ant-... GROQ_API_KEY=gsk_... DEEPSEEK_API_KEY=sk-... # Optional: multi-provider routing MULTI_PROVIDER=true BRAIN_PROVIDER=openai BRAIN_MODEL=gpt-4.1 WORKER_PROVIDER=groq WORKER_MODEL=llama-3.1-8b-instant CODER_PROVIDER=deepseek CODER_MODEL=deepseek-chat ANALYST_PROVIDER=openai ANALYST_MODEL=gpt-4.1-mini
# Start RedClaw as a systemd service sudo systemctl enable redclaw sudo systemctl start redclaw # Check status sudo systemctl status redclaw # View logs sudo journalctl -u redclaw -f
# Dashboard (web UI)
open http://localhost:8080
# API
curl -s http://localhost:8080/health
# Python SDK
python3 -c "
from redclaw.client import RedClawClient
c = RedClawClient('http://localhost:8080')
print(c.health())
"On Kali Linux, the agent has full system access: it can run any command, install tools, write files, and control the browser. Ensure you are running in an authorized environment.
RedClaw AI is designed for authorized penetration testing and security research only. Always obtain written permission before testing any system or network you do not own. Unauthorized use may violate computer crime laws.