agentguard

Scan codebases and evaluate runtime actions for security risks.

Updated Feb 16, 2026
One-click install
npx skills add https://github.com/rbkayz/gitskills --skill agentguard-rbkayz
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agentguard
Source: https://github.com/rbkayz/gitskills/tree/main/skills/registry/openclaw-agentguard
Command: npx skills add https://github.com/rbkayz/gitskills --skill agentguard-rbkayz

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill acts as a security guard for AI agents, preventing dangerous commands, data leaks, and the exposure of secrets by auditing actions and code.

Core Features & Use Cases

  • Security Auditing: Scans code and runtime actions for vulnerabilities and risks.
  • Action Evaluation: Determines if proposed actions (like network requests or command execution) are safe.
  • Trust Management: Manages and evaluates the trustworthiness of different skills.
  • Use Case: Before an agent executes a command to modify system files, AgentGuard can analyze the command and block it if it's deemed too risky.

Quick Start

Scan the current directory for security risks.

Frequently Asked Questions about agentguard

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

FAQPage Schema
How do I prevent AI agents from executing dangerous commands and leaking sensitive data?

To prevent dangerous commands and data leaks, AI agent security requires auditing runtime actions before execution. AgentGuard scans codebases and evaluates proposed actions like network requests or command execution, blocking operations deemed too risky based on predefined policies.

What is action evaluation in AI agent security and how does it work?

Action evaluation in AI agent security analyzes proposed runtime actions to determine if they are safe. It assesses risks like command execution, data exfiltration, and secret exposure, integrating with trust registries and predefined policies to block unsafe operations.

How do I scan my codebase for security risks before running an AI agent?

Scanning a codebase for security risks involves analyzing the current directory for vulnerabilities and dangerous patterns. AgentGuard performs this by evaluating code for command execution risks, potential data exfiltration, and secret exposure using predefined security policies.

Can I configure different protection levels for AI agent action auditing?

Yes, you can configure different protection levels for action auditing. AgentGuard supports the configuration of security levels and manages trustworthiness through trust registries, allowing you to adjust how strictly proposed agent actions are evaluated and blocked.

How do I manage trust for different skills used by my AI agent?

Managing trust for different skills involves evaluating and maintaining a trust registry for agent components. AgentGuard provides trust management features to assess the trustworthiness of various skills before allowing them to execute actions within your environment.