SkillSpector
A security scanner for AI agent skills used by Claude Code, Codex and other MCP-based tools, checking for prompt injection, data exfiltration, privilege escalation and other risks before a skill is installed.
SkillSpector scans AI agent skills, the kind used by Claude Code, Codex CLI, Gemini CLI and MCP servers, for vulnerabilities and malicious patterns before they are installed. In a research dataset of over 31,000 analyzed skills, the project found that 26.1% contained vulnerabilities and 5.2% showed likely malicious intent.
It checks 71 vulnerability patterns across 17 categories, including prompt injection, data exfiltration, privilege escalation, system prompt leakage, memory poisoning, tool misuse and MCP-specific risks such as least-privilege violations and tool poisoning. Analysis runs in two stages, a fast static pass and an optional LLM-based semantic evaluation, and it can look up live vulnerability data for known dependencies.
It accepts a Git repository, URL, zip file, directory or single file as input, and reports a 0-100 risk score in terminal, JSON, Markdown or SARIF format, with a baseline mechanism to suppress already-accepted findings on repeat scans. It is written in Python and is part of NVIDIA's Verified Skills pipeline, which scans, evaluates and signs skills before they are published to NVIDIA's skills catalog.