owasp-llm

Apply OWASP LLM Top 10 controls to LangChain workflows and agent tools.

Updated May 29, 2025

Search Tags:#owasp #langchain #llm #risk-management #ai-security #prompt-injection #agent-security

Ready to run

npx skills add https://github.com/crozzbite/DnDApp --skill owasp-llm

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: owasp-llm
Source: https://github.com/crozzbite/DnDApp/tree/main/.agent/skills/owasp-llm
Command: npx skills add https://github.com/crozzbite/DnDApp --skill owasp-llm

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

AI/LLM deployments face risks from prompt manipulation, data exposure, and unsafe tool usage. OWASP LLM Top 10 provides a structured, industry-aligned set of controls to mitigate these threats in AI workflows.

Core Features & Use Cases

Comprehensive coverage of OWASP LLM Top 10 controls tailored for LangChain flows, agent tools, and LLM outputs.
Practical patterns for prompt-injection detection, output sanitization, DoS protection, and secure tool usage.
Real-world use cases including securing chat agents, data pipelines, and automated decision systems.

Quick Start

Apply the OWASP LLM Top 10 controls to your LangChain prompts and agent tools to harden your AI applications.

Frequently Asked Questions about owasp-llm

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

FAQPage Schema

How do I prevent prompt injection in LangChain workflows?▼

To prevent prompt injection in LangChain workflows, apply OWASP LLM Top 10 controls that detect malicious patterns and sanitize inputs. This Skill implements threat modeling and pattern detection to secure chat agents and data pipelines against manipulation.

What are the OWASP LLM Top 10 controls for AI security?▼

OWASP LLM Top 10 controls for AI security are a structured set of mitigations for data exposure, DoS attacks, and unsafe tool usage. This Skill applies these controls across LLM outputs and agent tools to ensure secure automated decision systems.

Can I use this to sanitize LLM outputs in automated decision systems?▼

Yes, you can sanitize LLM outputs in automated decision systems using this Skill's output sanitization patterns. It applies OWASP LLM controls to prevent data exposure and misuse across agent tools and data pipelines.

How do I secure agent tools against misuse in AI applications?▼

You secure agent tools against misuse by applying OWASP LLM Top 10 controls for secure tool usage. This Skill provides practical patterns to mitigate risks in AI applications, protecting LangChain flows from unsafe tool execution.

Does this Skill work for threat modeling in AI data pipelines?▼

Yes, this Skill works for threat modeling in AI data pipelines by applying OWASP LLM Top 10 controls. It satisfies requirements for pattern detection and input sanitization to prevent data exposure and misuse in automated workflows.