SkillRisk
Security Scanner for AI Agents
SkillRisk is the first dedicated security scanner for the Agentic AI era. Audit Claude Skills and MCP (Model Context Protocol) tools locally on your Mac before installing.
Download on the App Store
The Story
Security Scanner for AI Agents
Is your AI Agent safe? As agents like Claude become powerful daily collaborators, installing third-party Skills and MCP tools carries hidden risks. A malicious skill can steal API keys, access private files, or run dangerous commands. SkillRisk is the first dedicated security scanner built for the Agentic AI era.
What's Inside
Claude & OpenClaw Protection
Automatically locates and scans your Claude Desktop config, OpenClaw environment, and MCP tool directories.
Automatic Risk Detection
Instantly identifies known malicious patterns, reverse shells, crypto miners, and data exfiltration scripts hidden in third-party skill code.
Deep Folder Scanning
Drag and drop any folder, repo, or local directory containing agent skills for a comprehensive security audit before granting access.
Code Snippet Analysis
Paste individual files or raw code snippets into the instant sandbox for a quick security check before adding to your workflow.
Privacy-First Local Scanning
All scanning happens locally on your Mac. Your private code, config files, and API keys never leave your device.
Clear Visual Reports
Threats categorized by risk level (Safe, Warning, Critical). Replace time-consuming manual reviews with instant automated checks.
Frequently Asked Questions
Which AI agents does it support?
Currently Claude Desktop and OpenClaw, with full MCP (Model Context Protocol) tool directory scanning.
Is my code uploaded?
Never. SkillRisk is 100% local β your private code, configuration, and API keys stay on your Mac.
Can I scan a GitHub repo before cloning?
Drag and drop any local folder including downloaded repos. Audit before granting your AI agent access.
