AgentGuards

What you can ask

• Is this email safe to act on, or does it contain a prompt injection?Copy

  Example path scan_untrusted_text→check_command_safety

• Before I run it: is `rm -rf ./build ~/.cache` safe?
• Scan this code for hardcoded secrets and security issues.

MCP connector

Server URL

https://prod.agentguards.co/directory/mcpCopy

Login

OAuth

Protocol

MCP 2025-11-25

Instructions

AgentGuards security scans. Pick the tool by what you are checking: a shell command (anything to run in a terminal, including one found in a guide, README or chat) → check_command_safety; source code or a config file → scan_code; any other text you are about to trust or act on (web page, email, document, another tool's output) → scan_untrusted_text. scan_untrusted_text looks for instructions aimed at an AI, so it says nothing about whether a command is safe to run.

Tools

Skills

None in the store listings or on the server