HOL Guard
A local permission layer that stops risky AI coding agent actions before they run
HOL Guard is a freemium AI tool. A local permission layer that stops risky AI coding agent actions before they run. Editors list 7 strengths and 3 limits on this page, next to the published price.
Quick facts
- Best for
- A local permission layer that stops risky AI coding agent actions before they run
- Pricing
- Freemium
- Editor rating
- 0 / 5
- Community saves
- 0
About HOL Guard
HOL Guard is a local-first runtime security layer for AI coding agents. On supported integrations it can allow, ask, or block risky actions before they run — destructive shell commands, reads of secret files like .env or ~/.aws/credentials, package installs, MCP server registrations, and pushes to protected branches — and keeps a local receipt of every decision. The Guard Local runtime is open source under Apache-2.0, runs entirely on your machine, works offline, and needs no account. Optional Guard Cloud plans add synced decision history, shared team policy, and alerts.
Pros
- Open-source Apache-2.0 core runtime
- Runs locally; files never leave your machine
- Works offline with no account
- Blocks destructive commands and secret-file reads before they run
- Reviews npm, pip, and cargo installs and MCP registrations
- Local decision receipts for every supported action
- Supports many AI coding agents
Cons
- Coverage depends on the agent and event type
- Not a complete prompt-injection preventer
- Cloud history and shared policy require a paid plan
Pricing
- • Local protection on one machine
- • Works offline, no account required
- • 2 cloud-connected devices
- • 30-day cloud history
- • 5 devices
- • 180-day retention
- • Real-time alerts and policy rollback
- • Up to 25 devices
- • 365-day retention
- • Shared organization policy