Free AI check and guard

Both tools run on your computer. You don’t need a Shield360 account.

Install the command line tool

curl -fsSL https://app.shield360.ai/api/cli/install-cli.sh | sh

Check your computer

shield360 ai-check

The report lists your AI apps, coding agents, editor extensions, local model servers and MCP servers, then shows what to fix and what to review:

Finding	What it means
Coding agent with nothing checking it	Its commands and file reads aren’t checked before they run
Local model server open to other computers	It accepts connections from beyond this computer
MCP server started through a package runner	npx, bunx, uvx or pipx can download new code when it starts
Remote MCP server	Your AI client sends tool calls and their data to that host

The check sends nothing. It reads names, versions, hosts and command names, and never reads an MCP server’s arguments, environment or tokens. It can’t tell http from https on a remote MCP server, or whether a package version is pinned.

Add --json for output you can use in a script.

Turn on the guard

shield360 coding guard on

This adds hooks to the coding agents on your computer and turns on blocking. It stops secrets and personal data in prompts and tool calls, and reads of secret files such as .env. Prompt injection gets a warning.

Command	What it does
shield360 coding guard on --warn	Warns instead of blocking
shield360 coding guard status	Shows whether each agent is protected
shield360 coding guard off	Turns the guard off

Blocked events are not reported anywhere when you have no account.

If your organization manages Shield360 on your computer, its policy stays in charge and these commands won’t change it.