OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Security & Sandboxing

Hol Guard

Hol Guard is an open-source Python CLI security layer for AI-agent workflows. It reviews shell commands, file and secret access, package installs, MCP tool calls, prompts and tool results, then allows, blocks or requests approval according to policy. It also inventories plugins, skills and agent configuration, and offers a separate plugin scanner for artifact checks.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Hol Guard — OpenViST project profile in Agent Security & Sandboxing.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Hol Guard when connecting supported AI agents to tools, files, packages or MCP servers and you want policy-based review before actions run. Use its plugin-scanner commands to inspect extension artifacts as a separate preparation step.

Who it is for

  • Developers integrating security checks into AI-agent tool execution.
  • Teams managing agent integrations, approvals and local decision history.
  • Plugin and skill maintainers who want to scan or lint artifacts before use.

Documented uses

  • Review risky shell commands, file access and package installation before execution.
  • Screen supported prompts and tool results for prompt injection and sensitive content.
  • Inspect MCP server configuration and tool calls through supported hooks and managed proxies.
  • Route decisions to native prompts or an approval center and retain local decision receipts.
  • Scan, lint, verify or prepare plugin submission artifacts with the related plugin-scanner CLI.
Before you choose

Requirements & limits

The repository documents local use without an account. Optional Guard Cloud provides shared history, team policy and fleet management. Installation requires Python 3.10 or newer and pipx; the project is open source under Apache-2.0.

  • Integration coverage varies with the events each agent exposes; the project provides a support matrix covering enforcement, approval delivery and failure behavior.
  • Guard connects through native agent hooks, managed MCP proxies and launch integrations; the relevant mechanism depends on the agent.
  • The local dashboard manages protection, approvals and decision history; cloud use is optional for shared history, team policy and fleet management.
  • The related plugin-scanner CLI covers artifact scanning and verification, distinct from runtime review of agent actions.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install and initialize Hol Guard

Install the CLI and start its first-run protection setup.

Before you start

  • Python 3.10 or newer
  • pipx

Steps

  1. Install the package with pipx install hol-guard.

  2. Run hol-guard init to start the setup wizard.

  3. Use the wizard to discover supported agents and configure protection, reviewing its prompts before setup changes.

Expected outcome

The CLI is installed and the wizard guides the initial protection setup for supported agents.

Guide limitations

  • The wizard only configures integrations for supported agents; coverage depends on events exposed by each integration.
  • The documented installation steps do not describe a completed agent run or verify a particular setup.
Documentation 1 ↗
Source-based practical guide

Set up and launch the Codex integration

Install Guard's Codex integration and use the documented dry-run and managed-launch commands.

Before you start

  • Hol Guard installed and initialized
  • Codex

Steps

  1. Run hol-guard install codex.

  2. Run hol-guard run codex --dry-run to record the current artifact state before launch.

  3. Run hol-guard run codex for a managed launch.

Expected outcome

The documented workflow installs Codex's native pre-tool hooks and provides a managed launch path; Guard refuses the managed launch if those hooks are missing or disabled.

Guide limitations

  • Integration coverage depends on the events exposed by Codex; consult the agent support matrix for coverage and failure behavior.
  • The commands are documented workflow steps, not a guarantee of compatibility with every environment.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
12 October 2026
Original website title
hashgraph-online/hol-guard · Official homepage title · retained in its source language
How we review sourcesSuggest a correction