AgentGuard

AI Agent Skill Scanner

Know what your AI agents are installing.

Scan AI agent Skills for security risks, hidden capabilities, and quality issues before they touch your codebase.

Scan a SkillSee how it works

Free to use · No account required · Analysis in seconds

SKILL.mdpreview
---
name: frontend-design
description: Build production-
  ready frontend interfaces
---

# Frontend Design

Read project files before
editing. Prefer semantic HTML.
Never touch .env secrets.

Security

92 / 100

Quality

88 / 100

Risk

LOW

Capabilities

  • Read project files
  • Execute shell commands
  • Modify source files
  • Network access

Scan before you install.

Paste a SKILL.md and get a security and quality assessment in seconds.

SKILL.md0 chars

Supported: SKILL.md · Max 80,000 chars

How it works

A short path from Skill text to an actionable trust assessment.

01

Paste

Paste your SKILL.md or Skill instructions into the scanner.

02

Analyze

AgentGuard checks capabilities, dangerous instructions, suspicious behavior, and quality.

03

Understand

Get a clear Trust Report explaining what the Skill can actually do.

What we analyze

AgentGuard combines deterministic pattern checks with AI-assisted analysis to surface capabilities and risks before installation.

  • Shell commands
  • File system access
  • Environment variables
  • Network requests
  • External downloads
  • Credential access
  • Suspicious URLs
  • Obfuscated instructions
  • Prompt injection patterns
  • Dangerous / destructive operations

Example security findings

Illustrative findings — your report only includes issues evidenced in the Skill you scan.

LOW

Shell command execution detected

The Skill instructs the agent to execute shell commands during setup.

Why this matters: Shell commands can modify files or execute programs on your machine.

Recommendation: Review the commands before installation.

MEDIUM

External network request detected

The Skill references an external URL that could be fetched at runtime.

Why this matters: Network access can leak context or pull untrusted content.

Recommendation: Confirm the destination and whether the call is necessary.

FAQ

Don't install blindly.

Understand what your AI agent Skills can do before you trust them.

Scan a Skill — It's Free