skill-scanner
AI Agent skill security scanner that detects injections and malicious code.
- Type
- Agent Skill
- Open source
- Yes
- GitHub Stars
- ★ 2.2k
- Source
- skill-github
Overview
Skill Scanner is a security scanning tool for AI Agent skills, capable of detecting prompt injection, data leakage, and malicious code patterns. It combines pattern-based detection (YAML + YARA), LLM as a judgment engine, and behavioral data flow analysis to maximize threat detection coverage while minimizing false positives. Supports OpenAI Codex Skills and Cursor Agent Skills formats, and can be extended with a plugin architecture for custom analyzers. Ideal for developers and teams requiring security reviews of AI skills.
Capabilities
- ▪Detect prompt injection
- ▪Detect data leakage
- ▪Detect malicious code patterns
- ▪Multi-engine detection
- ▪False positive filtering
- ▪CI/CD integration
Use cases
Setup
Using uv (recommended): `uv pip install cisco-ai-skill-scanner` or using pip: `pip install cisco-ai-skill-scanner`
This information was compiled by AI from public sources and may contain inaccuracies — please refer to the source.
FAQ
Does a clean scan result mean the skill is secure?
No, it means no known threats were detected, but absolute security cannot be guaranteed.
How to handle false positives and false negatives?
Adjust scanning strategies to balance between false positives and false negatives.
Related skills
gemini-cli
Gemini CLI is an open-source AI agent that brings Gemini's powerful capabilities directly into the terminal.
private-gpt
Provides a complete API layer for local models, supporting RAG, skills, tools, and more.
planning-with-files
File-based persistent planning skill, ideal for AI coding agents and long-running tasks.
scientific-agent-skills
Transform any AI agent into a scientific assistant with 147 ready-to-use research skills.
susi_alexa_skill
A skill that enables question-and-answer interactions between Alexa and Susi AI.
claude-context
Provides code search capabilities for Claude Code, turning the entire codebase into context.