Your AI agent can run commands. It should not get a free pass. HOL Guard is an open-source, local-first security layer for builders running AI agents, tools, plugins, skills, MCP servers, and package…

HOL Guard reviews agent tool calls and holds risky ones for approval. v3.0.182-189: secret reads gate and queue for approval, approvals bind to one action, shell parser gaps closed, OpenCode review…

HOL Guard: Block AI threats at runtime with Python's open-source antivirus. See link below 👇

The market context: • Uber: 5K engineers on Claude Code. Blew their 2026 AI budget in 4 months. 11% of deploys are now agent-only, no human in the loop. • NVIDIA: 30K devs on Cursor. 3x more…

Secure your AI agents with hol-guard, an AI antivirus for tools like Claude Code and Cursor. It inspects MCP servers and plugins before execution to prevent malicious actions. Essential for the Model…

做 Agent 最怕的场景，是它把危险命令当正常步骤执行，HOL Guard 就是冲着这个问题来的。 GitHub： 官网： 给 Codex、Claude Code、Cursor、Gemini、OpenCode 这类开发 Agent 加一层安全拦截，在工具执行前识别危险提示、拦截高风险操作，支持 4 档保护级别和本地审批中心。 适用场景： • 已经让 Agent…

AI antivirus for developer agents: protect Codex, Claude Code, Cursor, Gemini, OpenCode, plugins, skills, MCP servers, and AI harnesses before tools run.

Protect your harness locally with hol-guard. Use plugin-scanner when you need maintainer and CI checks for plugins, skills, MCP servers, and marketplace packages.

If this works on messy repos, it is more useful than half the polished AI demos on launch day. AI antivirus for developer agents: protect Codex, Claude Code, Cursor, Gemini, OpenCode, plugins,…