# Guardana Control MCP server

> Supervise AI agent runs by checking recorded actions against policies and enforcing approvals or vetoes.

- Repository: https://github.com/guardana/control
- Page: https://hostedmcp.org/servers/guardana/control
- Transport: —
- Stars: 11
- License: Apache-2.0

## FAQ

### Can I use this as a security boundary?

No, the evidence trail is not tamper-evident, so you should not deploy this as a security boundary.

### How does the MCP gateway handle tool calls?

The MCP gateway consults signed policies before a tool call and can be configured to block tools whose definitions no longer match their reviewed fingerprint.

### What happens if the external decision point is silent?

If the external PDP is silent, it blocks the calls it governs when operating in an enforcing mode.

### Does the system store the content of the agent's reasoning?

No, content capture is disabled by default and hidden reasoning is never stored.