Show HN: Ctrl – execution control plane for AI agents (github.com)

Agents are starting to take real actions (publish content, trigger workflows, touch prod systems). Ctrl is an execution control plane that sits between agent intent and real-world actions. It intercepts MCP tool calls, scores risk, enforces allow/deny/approve policies, and executes only what’s authorized with a full SQLite audit ledger. It Ships as a drop-in CtrlMCP wrapper for LangChain. Looking for feedback and early pilots running agents in production.