skills.rest
SensLiao
SensLiaoCOMMUNITY
Explore all skills by this author →
GitHub

security-app-llm

Enforces OWASP LLM Top 10 security standards on GenAI applications.

Updated May 28, 2026
Search Tags:#AI security#LLM security#security compliance#Agentic AI#GenAI security
One-click install
npx skills add https://github.com/SensLiao/Claude-code-setting --skill security-app-llm
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: security-app-llm
Source: https://github.com/SensLiao/Claude-code-setting/tree/main/skills/security-app-llm
Command: npx skills add https://github.com/SensLiao/Claude-code-setting --skill security-app-llm

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill addresses the security concerns in GenAI, LLM, and Agentic AI applications, ensuring that they are protected against various threats and vulnerabilities.

Core Features & Use Cases

  • Comprehensive Security Checks: Covers OWASP LLM Top 10, Agentic AI threats, and other relevant standards.
  • Tool Permission Boundaries: Ensures that agents have the correct permissions to avoid misuse.
  • Memory and Context Poisoning Protection: Prevents malicious instructions from being executed.
  • Prompt Injection Defense: Protects against direct and indirect prompt injection attacks.
  • Output Validation: Ensures that LLM outputs are safe and do not contain vulnerabilities.
  • Human Override and Rollback: Allows users to stop or undo agent actions if needed.
  • Model and Data Provenance: Ensures that the model and data sources are trustworthy.
  • Compliance with Standards: Adheres to OWASP, NIST, MITRE, and other relevant standards.
  • Use Case: Ideal for securing AI applications in industries like finance, healthcare, and government.

Quick Start

Use the security-app-llm skill to review the security of your AI application.

Frequently Asked Questions about security-app-llm

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I secure LLM applications against OWASP Top 10 vulnerabilities?▼

Securing LLM applications against OWASP Top 10 vulnerabilities requires enforcing security standards, checking for vulnerabilities, and providing controls to mitigate risks. This includes defending against prompt injection and ensuring output validation.

What is the best way to prevent prompt injection in GenAI applications?▼

Preventing prompt injection in GenAI applications involves implementing specific defense mechanisms that protect against direct and indirect injection attacks. Additionally, you must ensure memory and context poisoning protection to prevent malicious instructions from executing.

How do I enforce tool permission boundaries for Agentic AI?▼

Enforcing tool permission boundaries for Agentic AI ensures that agents have the correct permissions to avoid misuse. You should also implement human override and rollback capabilities to stop or undo agent actions if needed.

Does this approach secure AI applications for compliance in finance and healthcare?▼

Yes, securing AI applications for compliance in finance and healthcare requires adherence to OWASP, NIST, and MITRE standards. This approach ensures data protection and mitigates threats specific to these highly regulated industries.

How do I validate LLM outputs to ensure they do not contain vulnerabilities?▼

Validating LLM outputs to ensure they do not contain vulnerabilities requires implementing output validation checks. This ensures that generated content is safe and compliant with established security frameworks before being delivered to end users.

Related Skills

balic-AI-ML-R-D-Resourcesbalic-AI-ML-R-D-Resources

security-audit-context-building

Map architecture, trust boundaries, and data flows for security audit scoping.

Official
Advanced
gparker97gparker97
4

deploy-prod

Commit, push, monitor CI, and deploy Vue and Astro production builds.

Community
Advanced
tomas-utomas-u
1

security

Guide security architecture design with STRIDE threat modeling and compliance assessments.

Community
Advanced
EntroVyxEntroVyx
3

llm-prompt-injection

Test LLM endpoints for prompt injection vulnerabilities using Python scripts.

Community
Advanced