> ## Documentation Index
> Fetch the complete documentation index at: https://docs.civic.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Guardrail Proxy

> Security policies that control what AI assistants can access and how they process data

**Flask status: Graduated to Civic**

<Note>
  **This feature has graduated!** Guardrails are now a core feature of Civic. For the complete documentation, see [Guardrails in Civic](/civic/concepts/guardrails).
</Note>

## Overview

The Guardrail Proxy started as a labs project to implement security rules between AI assistants and MCP tools. It has since become a core feature of Civic, available to all users.

## What's available in Civic

* **Conversational management**: Add, list, and remove guardrails through natural language
* **Request guardrails**: Block or validate parameters before tools execute
* **Response guardrails**: Redact PII, transform data, and filter outputs
* **Built-in protection**: Prompt injection detection, PII patterns, and unsafe file blocking
* **Hierarchy controls**: Account, toolkit, and user-level policies

## Get started

See the [Guardrails documentation](/civic/concepts/guardrails) to:

* Learn how guardrails work
* View available guardrail templates
* Add guardrails to your toolkits
* Configure PII redaction and other protections

## Integration

Guardrails integrate with all Civic features:

* **[MCP Hub](/labs/projects/mcp-hub)**: Centralized management for all your MCP servers
* **[Bodyguard](/labs/projects/bodyguard)**: Additional prompt-level threat detection