IronCurtain
A secure runtime for AI agents
$
npx @provos/ironcurtain architecture
Agent (LLM)
Claude · GPT · Gemini
V8 Sandbox
Code runs in isolated runtime
Trusted Process
Policy engine · allow/deny/escalate
MCP Servers
filesystem · git · fetch · sandboxed via Anthropic SRT
- Write security policy in plain English - a constitution, not a DSL
- Every MCP tool call evaluated against compiled policy rules
- V8 sandbox isolation - agents never touch your filesystem directly
- Human-in-the-loop escalation for sensitive operations
- Agent-framework agnostic - works with any agent that speaks MCP