A new tool lets teams apply real-time guardrails to Claude Code tool calls using cloud-based decision tables. By hooking into Claude’s PreToolUse event, the system enables instant policy enforcement without code changes or restarts.

The approach allows fine-grained control over shell commands, file operations, and MCP tools. Teams can block risky actions like rm -rf or kubectl delete while allowing exceptions. Rules update immediately across all users and sessions.

This method offers a lightweight way to enforce security policies in dynamic AI-assisted development environments. It provides visibility into tool usage and blocking patterns. The team can self-host and redact sensitive data before it leaves their infrastructure if needed.