hol-guard

hol-guard

Open-source antivirus protection for AI agents, blocking risky tools and malicious behaviors.

MCPDevOpen source
Type
MCP
Open source
Yes
GitHub Stars
★ 1.0k
Source
mcp-github

Overview

HOL Guard is an open-source AI agent antivirus solution that blocks risky tools, secret access, prompt injection, malicious packages, MCP servers, plugins, and skills at runtime. It detects and prevents potential security threats by reviewing agent operations such as shell commands, file access, package installations, and MCP tool calls. HOL Guard runs locally without requiring an account and provides a CLI and local dashboard for managing protection policies, resolving approval issues, and reviewing decision history. Optional Guard Cloud offers shared history, team policies, and device management features.

Capabilities

  • ▪Blocks risky tools and malicious packages
  • ▪Prevents secret exposure and destructive operations
  • ▪Detects prompt injection and supply chain risks
  • ▪Supports multiple AI agents and MCP servers

Use cases

Protect AI agents from malicious attacksEnsure AI operations comply with security policiesManage security policies across multi-agent environments

Setup

Requires: Python 3.10+pipx
pipx install hol-guard && hol-guard init

This information was compiled by AI from public sources and may contain inaccuracies — please refer to the source.

FAQ

Which AI agents does HOL Guard support?

Supports multiple AI agents including Codex, Claude Code, GitHub Copilot CLI, and more.

How do I update HOL Guard?

Use the command `hol-guard update` to update your existing installation.

Related skills