agentguard

Block dangerous AI agent commands and prevent data leaks.

454|66|Updated Feb 2, 2026

Search Tags:#security #audit-logging #trust-registry #runtime-protection #GoPlus #agentguard

Ready to run

npx skills add https://github.com/GoPlusSecurity/agentguard --skill agentguard

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agentguard
Source: https://github.com/GoPlusSecurity/agentguard/tree/main/skills/agentguard
Command: npx skills add https://github.com/GoPlusSecurity/agentguard --skill agentguard

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires @goplus/agentguard, and includes scripts (resource) components.

What problem does it solve?

GoPlus AgentGuard provides real-time security protections for AI agents by automatically blocking dangerous commands, preventing data leaks, and tracking which skill initiated an action. It reduces risk when integrating third-party skills and prompt injections.

Core Features & Use Cases

Real-time protection: blocks high-risk commands and sensitive actions during agent operation.
Skill governance: integrates with a trust registry to attest, look up, and enforce per-skill policies.
Auditing & reporting: logs security events and traces action origins for accountability.
Use Case: Audit a new skill before deployment and auto-scan for safety risks across code, data, and configurations.

Quick Start

Install via npm: npm install @goplus/agentguard
Start a session to enable automatic scanning and protection for new skills
Use subcommands (scan, action, trust, report, config) to manage protections

Frequently Asked Questions about agentguard

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

FAQPage Schema

How do I block dangerous commands and prevent data leaks from AI agents?▼

To block dangerous commands and prevent data leaks, this Skill applies real-time runtime protection by evaluating actions and stopping high-risk operations before execution. It tracks the originating skill to ensure accountability.

What is a trust registry and how does it manage AI agent security policies?▼

A trust registry manages AI agent security by attesting, looking up, and enforcing per-skill rules. It integrates with the runtime engine to govern which third-party skills are allowed to execute specific actions safely.

How do I audit a new skill for security risks before deployment?▼

To audit a new skill for security risks, use the scan subcommand to automatically evaluate code, data, and configurations. This generates a detailed security report highlighting potential prompt injection or dangerous action vulnerabilities.

Do I need to install npm packages to use runtime protection for AI agents?▼

Yes, installing the @goplus/agentguard npm package is required to enable runtime protection for AI agents. This dependency provides the core engine required for scanning, action evaluation, and trust management workflows.

Can I trace which skill initiated a blocked action in my security logs?▼

Yes, you can trace which skill initiated a blocked action by querying the audit logs. The system logs security events and traces action origins across modules to provide full accountability for runtime protections.