AgentGuard
Scan AI agent Skills for security risks before you install them.
Last verified:
What is AgentGuard?
AgentGuard is a free AI agent Skill scanner that analyzes SKILL.md files for security risks, hidden capabilities, and quality issues before installation. It targets developers using AI coding agents who want to understand what agent Skills can do — such as shell execution, file access, or network requests — before trusting them in their codebase.
AgentGuard pricing
Pricing model: Free
Free to use, no account required
AgentGuard pros
- Fast analysis in seconds with no account required
- Combines deterministic pattern checks with AI-assisted analysis
- Covers a broad range of risks including prompt injection, credential access, and obfuscated instructions
- Produces a clear Trust Report with security, quality, and risk scores plus actionable recommendations
AgentGuard cons
- Limited to SKILL.md files up to 80,000 characters
- Does not guarantee a Skill is safe — findings are advisory only
- No visible team, company, or pricing/enterprise information disclosed
Frequently asked questions about AgentGuard
Is AgentGuard free?
Yes, AgentGuard is free to use and requires no account.
How does AgentGuard work?
You paste a SKILL.md file, AgentGuard analyzes capabilities, dangerous instructions, suspicious behavior, and quality, then returns a Trust Report.
What kinds of risks does AgentGuard detect?
It detects shell commands, file system access, environment variable access, network requests, external downloads, credential access, suspicious URLs, obfuscated instructions, prompt injection patterns, and destructive operations.
Does AgentGuard guarantee a Skill is safe?
No — it provides an assessment, and users should still review findings before installation.