AI tools directory / Coding

HOL Guard

Approve or block risky AI coding agent actions locally.

Checked against hol.org on Nov 11, 2026

Visit hol.org

What HOL Guard does

HOL Guard is a security layer that runs on your own computer and sits between AI coding agents and the actions they try to take. Before an agent runs a destructive command, reads secrets, installs software or moves data, Guard can allow it, ask you, block it or just log it. It is made by HOL, a group working on standards for AI agents. The local runtime is open source under Apache-2.0, and an optional paid cloud service adds synced history and team policy.

Best for

  • Developers running Claude Code, Codex or Cursor on real projects
  • People who want to approve risky agent commands before they run
  • Teams that need a shared policy and audit trail for AI agents

Key features

Pre-action checkpoints

Agent actions on supported integrations can be allowed, sent to you for approval, blocked or only observed.

Local receipts and audit log

Guard keeps a local record of what each agent did and what was decided, and works offline.

Config and package scanning

Checks MCP configurations, skills, plugins and packages for risky or changed code.

Secret and injection protection

Targets credential exposure and actions triggered by prompt injection.

Wide agent support

Works with Codex, Claude Code, Cursor, OpenCode, GitHub Copilot CLI, Kimi, Grok and several others.

CI scanner

A plugin scanner is available through pipx and as a GitHub Action for checking plugins in CI.

HOL Guard pricing

HOL Guard has a free plan. Guard Local is free with no account or card: local scanning, command interception, approvals and a local audit log on one machine. Paid plans start at $4.99/mo. Free trial of Pro (length shown at checkout). An annual toggle says 2 months free, but annual prices are not listed. Pro and Team need a payment method at activation. Prices checked on the official site on Nov 11, 2026.

Free$0One machine, local only
Solo$4.99/moCloud sync for 2 devices, 30-day history
Pro$15/mo5 devices, 180-day history, real-time alerts
Team$30/seat/moUp to 25 devices, 365-day history, shared policy
EnterpriseContact salesContact sales

Source: HOL Guard pricing page

Pros and cons

What it does well

  • The core enforcement runs locally and is open source, so it works without a cloud account.
  • Supports a long list of coding agents rather than a single one.
  • The free tier already covers approvals, blocking and a local audit log.

Know before you sign up

  • How much Guard can intercept depends on the agent and the type of event; HOL publishes a coverage matrix.
  • The free plan is limited to one machine; cross-device history needs a paid plan.
  • Cloud history is kept for 30 to 365 days depending on the plan.

Alternatives to HOL Guard

Questions about HOL Guard

What is HOL Guard?

HOL Guard is an AI tool from hol.org, listed in our Coding category. Approve or block risky AI coding agent actions locally.

Is HOL Guard free?

Yes, HOL Guard has a free plan, and paid plans start at $4.99/mo.

How much does HOL Guard cost?

Paid plans start at $4.99/mo, as listed on hol.org on Nov 11, 2026.

What are some alternatives to HOL Guard?

Tools that do a similar job include Qodo.