maxwellyoung

Skillscan — Security skill for Claude Code

Security community

Static security scanner for Claude Code skills and MCP servers.

How to install Skillscan

This entry records only its repository, not the path inside it, so there is no exact command to give. Open maxwellyoung/skillscan and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Skillscan does

Static security scanner for Claude Code skills and MCP servers.

Alternatives in Security

Skill Scanner — Security Scanner for Agent Skills 2.5k ★
Mcp-scan (Invariant Labs) — MCP security scanner with proxy mode for real-time scanning without infrastructure changes 1.9k ★
Promptmap — Security scanner for custom LLM apps 1.2k ★

README

SkillScan

Static security scanning for AI skills, MCP servers, npm packages, VS Code extensions, and GitHub Actions workflows. Use it from the web UI, the `POST /api/scan` API, the `skillscan` CLI, or the GitHub Action (JSON and SARIF output).

SkillScan checks local skill instructions, GitHub-hosted code, npm package metadata, OpenVSX extension metadata, GitHub Actions workflows, and MCP server repositories for risky patterns before you install or run them. It is intentionally fast and boring: static analysis, clear findings, and remediation notes.

The ClawdHub Supply Chain Crisis

Nick Saraev's viral "It Got Worse - Clawdbot" video exposed:

No vetting process on ClawdHub - malicious skills published freely
Thousands of compromised Clawdbot instances via Shodan scanning
API token theft through malicious skill instructions
Supply chain attacks via skill repositories

His advice? "Read every file or feed files to AI to check safety." **We automate that.**

What it demonstrates

Static analysis for a fast-moving AI tooling ecosystem.
GitHub URL, npm package URL, OpenVSX extension URL, file, and direct-code scanning.
Risk scoring with severity, category, location, snippet, and remediation.
A focused security UI that avoids hiding findings behind generic AI summaries.
A public example of building small developer tools with practical failure modes.

Features

ClawdHub and Claude Code skill attack detection.
MCP server and GitHub repository scanning.
npm package metadata and lifecycle-script scanning.
VS Code/OpenVSX extension manifest and source-behavior scanning.
GitHub Actions workflow scanning for release-pipeline and PR-trigger risks.
0-100 security score with A-F grades.
Pattern checks for shell execution, network access, file-system access, prompt injection, credential patterns, data exfiltration, package install hooks, extension activation behavior, and CI workflow trust boundaries.
No AI inference req