HOL Guard

HOL Guard is an open-source, local-first security layer for AI agents and the tools, plugins, skills and MCP servers they use. It reviews what an agent is about to do before it runs and stops risky actions. It works with Codex, Claude Code, GitHub Copilot CLI, Cursor, Cline, Gemini CLI, OpenCode, OpenClaw and many other agents, and runs locally without an account.

HOL Guard checks shell commands, file access, package installs and MCP tool calls, and detects secret exposure, destructive operations, prompt injection and supply-chain risks. It can allow, block or ask for your approval, and a CLI and local dashboard let you manage protection, resolve approvals and review decision history. An optional Guard Cloud adds shared history and team policies, and a separate plugin scanner checks agent plugins, skills and MCP integrations before you publish them.

HOL Guard puts a safety net under AI coding agents that run with broad access to your machine, without sending your activity anywhere. It is Apache-licensed.

Stars

0

Forks

0

Commits

0

Repository

hashgraph-online/hol-guard

• SnykCommercial security platform for code and dependencies.
• NemoClawNo. 105Open-source secure sandbox for running AI agents.