Cisco Skill Scanner
Open-source scanner for AI agent skills and tool-use security.
What it does
An open-source tool from Cisco AI Defense for scanning AI agent skills and tools for security vulnerabilities. Analyses tool definitions, permission scopes, and input validation to identify risks in how agents interact with external systems.
Security relevance
Agent skills (tool use) are a primary attack surface — improperly configured tools can enable data exfiltration, privilege escalation, and unintended actions. This scanner checks whether skills have proper input validation, permission boundaries, and error handling before deployment.
When to use it
Use during development and before deployment of AI agents with tool-use capabilities. Integrates into CI/CD pipelines for continuous scanning. Lower barrier than the A2A Scanner — suitable for teams getting started with agent security testing.
OWASP coverage
Risks addressed — mapped to both OWASP Top 10 standards. 2 in LLM, 2 in Agentic.
The raw record
What Yuntona stores. Single source of truth — fork it on GitHub.
name: Cisco Skill Scanner slug: cisco-skill-scanner type: Mixed category: AI Red Teaming url: https://github.com/cisco-ai-defense/skill-scanner reviewed: 2026-04 added: 2026-04 updated: 2026-04 risks: llm: [LLM01, LLM07] asi: [ASI02, ASI04] complexity: Guided Setup pricing: — audience: Red Team lifecycle: [develop] tags: [Agentic, Cisco, Open Source, Scanner]