Guides: securing AI coding agents

Last updated 23 October 2026

Practical, plain advice for anyone who lets an AI agent work on their computer. Each guide starts with what you can do yourself, then shows what AgentWarden adds.

• What is antivirus for AI agents?
  Why AI coding agents need their own security layer, what antivirus for AI agents watches, and what it cannot do.
• How to secure Claude Code on your computer
  Practical steps to run Claude Code safely: permissions, sandboxing, secrets, MCP servers, backups, and how AgentWarden watches every step.
• How to secure Codex, Cursor and other AI coding agents
  A checklist for running Codex, Cursor, Gemini CLI, GitHub Copilot and other AI coding agents safely on Mac and Windows.
• Detecting prompt injection in AI coding agents
  How indirect prompt injection hijacks AI coding agents through web pages, issues, pull requests and tool results, the warning signs, and how to catch it.
• Checking MCP servers and agent skills for malicious instructions
  Why MCP servers, skills and plugins are a supply-chain risk for AI agents, what to check before installing one, and how AgentWarden scans them.