llm-security - LLM and AI Agent Security Assessment
Conduct authorized security assessments of LLM applications and AI agents, covering prompt injection, tool abuse, RAG exposure, and supply-chain risks.
Tags
Updated: 2026-09-16Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Reconnoiter AI attack surfaces
- Test prompt injection vulnerabilities
- Test tool abuse risks
- Test memory poisoning
- Test output security vulnerabilities
- Extract system prompts
Inputs
- Precedent pentest log
- Tool index configuration
- Skill supply chain guide
Outputs
- Reproducible assessment evidence
- Security assessment report
- Checklist verification status
Requirements
- Authorized assessment permissions
- Python runtime environment
- Node.js runtime environment