openguardrails

Sanitizes sensitive data and detects prompt injection before LLM processing.

18|4|Updated Jun 30, 2026
One-click install
npx skills add https://github.com/openguardrails/openguardrails --skill openguardrails
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: openguardrails
Source: https://github.com/openguardrails/openguardrails/tree/main/openclaw-security/clawhub-skill
Command: npx skills add https://github.com/openguardrails/openguardrails --skill openguardrails

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill protects AI agents from prompt injection, data leakage, and misuse by scanning inputs and outputs through a configurable detection pipeline.

Core Features & Use Cases

  • AI Security Gateway: Sanitizes sensitive data (PII, secrets) locally before sending to LLMs.
  • Prompt Injection Detection: Identifies and blocks malicious instructions embedded in content.
  • Use Case: When your AI agent processes an email, this Skill automatically redacts credit card numbers and API keys, then checks for hidden malicious commands before the agent acts on the email's content.

Quick Start

Install the OpenGuardrails plugin using the command openclaw plugins install @openguardrails/openclaw-security.

Frequently Asked Questions about openguardrails

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How does prompt injection detection work for AI agents?

Prompt injection detection works by scanning inputs and outputs through a configurable pipeline to identify and block malicious instructions embedded in content before it reaches LLMs. It analyzes content locally to prevent hidden commands from manipulating your AI agents.

How do I sanitize sensitive data like PII before sending to an LLM?

To sanitize sensitive data before sending to an LLM, use a security gateway that locally redacts PII and API keys from inputs. This ensures credit card numbers and secrets are removed from content like emails before the AI agent processes it.

Can I configure security policies for different LLM providers?

Yes, you can configure security policies across various LLM providers. The gateway sanitizes data and detects malicious instructions through a configurable detection pipeline, allowing you to apply tailored security rules for your specific AI agent workflows.

What's the best way to secure AI agents against data leakage and misuse?

Securing AI agents against data leakage and misuse is best achieved by deploying a security gateway that scans both inputs and outputs. This approach automatically redacts secrets and blocks malicious instructions, protecting agents from prompt injection threats.

Do I need to install a plugin to enable LLM security for my agents?

Yes, enabling LLM security requires installing the OpenGuardrails plugin via the command `openclaw plugins install @openguardrails/openclaw-security`. This sets up the local detection pipeline to sanitize data and block prompt injections.