HyperClaw
LLM gateway with per-tenant agent registry and tool allow-lists.
Architecture
MCP-native.
Container-isolated.
Not a chat wrapper — each scenario is an isolated agent with dedicated tools and policies.
How components connect
ClawDesk Client Electron · React · streaming workbenches
HyperClaw Gateway LLM routing · agent registry · auth
SkillGuard
AgentSec
Alert Triage
Phishing IR
Auto-IR
Vuln Scan
One isolated container per scenario · dedicated SOUL & tool scope
ClawDesk MCP triage · forensics · skillguard · agentsec · vuln_scan
Four building blocks
Scenario runtime
One container per scenario — separate SOUL, config, workspace.
ClawDesk MCP
Modular tools returning structured JSON for UI and audit.
Desktop client
Electron + React with streaming tool output and workbenches.
Scenario map
| Scenario | Category | Container |
|---|---|---|
| SkillGuard | AI Security | skillguard |
| AgentSec | AI Security | agentsec |
| Alert Triage | SOC | alert-triage |
| Phishing IR | SOC | phishing-ir |
| Auto-IR | SOC | auto-ir |
| Vuln Scan | SOC | vuln-scan |
Ready to see ClawDesk in action?