HOL Guard

What is HOL Guard?

AI coding agents such as Claude Code, Codex and Cursor can run terminal commands, open secret files, install packages and call MCP tools on your laptop. HOL Guard adds a local review step in front of those actions. Routine work passes through; risky actions are held so you can approve or block them on your own machine, and each decision is recorded locally. The core is Apache-2.0 open-source software installed as a Python CLI; Guard Cloud is an optional paid add-on, so pricing is freemium. HOL Guard is not a gateway, not a network firewall and not a complete prompt-injection defense, and its scans are not a safety guarantee. Install: pipx install hol-guard && hol-guard init

HOL Guard features

• Approve or block risky shell commands from coding agents
• Checks secret and credential file access before it happens
• Reviews package, plugin and skill installs before they run
• Checks MCP and tool calls from Claude Code, Codex and Cursor
• Keeps a local log of what was allowed or blocked
• Apache-2.0 open-source core with optional paid Guard Cloud

Who uses HOL Guard and for what

• Developers using Claude Code who want to approve risky shell commands
• Keeping coding agents from reading .env and credential files unasked
• Reviewing package and plugin installs an agent proposes
• Checking MCP servers and tool calls before an agent uses them
• Security-minded teams trying coding agents on developer laptops

HOL Guard FAQ

Comments (0)

No comments yet. Ask the maker a question or share how you would use it.