Launching today

AI Skill Spector
Scan AI agent skills before you install them
0 followers
Scan AI agent skills before you install them
0 followers
AI Skill Spector is a free, browser-based security scanner for AI Agent Skills and MCP configurations. Scan SKILL.md files, GitHub repositories, ZIP packages, and local files for prompt injection, suspicious commands, credential exposure, and potential security risks. Default static analysis runs locally in your browser, with no signup required.



Free
Launch Team / Built With

AWS Startup AdvisorPersonalized AWS guidance that meets you where you’re already building
Promoted



Hey Product Hunt! 👋
I'm the maker behind AI Skill Spector, built as part of KitVerse.
As AI coding agents become more capable, we're increasingly installing community-built Skills and connecting MCP tools to our development environments.
But there's a security question I think deserves more attention:
What should we check before trusting a third-party AI Agent Skill?
A Skill may contain more than just Markdown instructions. It can include shell scripts, Python code, configuration files, dependencies, and instructions that influence an agent's behavior.
I built AI Skill Spector to make that initial security inspection easier.
A few things I focused on:
Browser-based static analysis
Local-first processing by default
No registration required
GitHub, local files, ZIP, and text input
File-level security findings
JSON, Markdown, and SARIF report exports
I also wanted the default scanning workflow to avoid uploading Skill contents to another server.
This is an evolving project, and I'd really appreciate feedback from developers working with AI agents, Claude Code Skills, and MCP integrations.
A few questions I'd love your thoughts on:
How do you currently inspect third-party Skills before installing them?
Which security patterns do you think a static scanner should detect?
What would make a security report more useful in your workflow?
Give it a try:
https://kitverse.dev/ai-tools/skill-security-scanner/
Thanks for checking it out! 🚀