HOL Guard

What it is

HOL Guard provides AI agent runtime security through two complementary layers. Guard Local is an open-source runtime that runs on your machine, evaluating supported agent actions before execution and applying configured policy to allow, observe, ask, or block actions. It protects against secret exposure, destructive commands, prompt-injection-driven actions, malicious packages, risky MCP configuration, and unsafe plugins. Guard Cloud is an optional service that adds synchronized approval history, shared policy, team dashboards, and fleet visibility without requiring cloud connectivity for local enforcement.

The product works by intercepting supported agent events across tools like Cursor, Claude Code, Codex, OpenCode, and Antigravity CLI. Guard can stop actions before they run, pause for human approval, or allow them through based on policy. All decisions are recorded locally in an audit trail. For teams, shared policies ensure consistent enforcement across developers while individual machines maintain local decision authority.

Quick answers

What is HOL Guard?

HOL Guard is a local-first runtime security layer for AI agents. Its open-source Guard Local runtime evaluates supported tool calls and local artifacts, blocks known threats, pauses ambiguous actions for approval, and records security receipts. It runs on the developer machine without a cloud account, while Guard Cloud is an optional service for synchronization, shared policy, fleet visibility, and team workflows.

What does HOL Guard protect against?

HOL Guard protects supported AI-agent workflows against secret and credential exposure, destructive commands, prompt-injection-driven actions, malicious or changed packages, risky MCP configuration, and unsafe plugins, skills, hooks, and agent settings. Enforcement depth varies by agent and event type, with a public coverage matrix showing what is blocked, reviewed, observed, or not covered.

Is HOL Guard open source?

Yes. HOL Guard's core Guard Local runtime is open source under Apache-2.0. It can be installed, inspected, and used without a paid plan or cloud account, and normal local protection continues offline. Guard Cloud is an optional, separately scoped service for synchronized evidence, shared policy, fleet visibility, and team workflows.

Do I need Guard Cloud to block destructive commands?

No. Guard Local intercepts actions and enforces policy on your machine without a cloud connection. Guard Cloud adds shared history, dashboards, and team workflows, but local blocking does not depend on it.

Does Guard upload source code or secrets?

No for Guard Local. It evaluates actions on the machine. If you enable Guard Cloud, sync stores redacted decision summaries, not your source code or secret material.