skills.rest
roanbrasil
roanbrasilCOMMUNITY
Explore all skills by this author →
Website𝕏@roanbrasilGitHub

ai-safety-guardrails

Implement input validation, prompt injection defense, and output moderation for LLM applications.

5|1|Updated Jun 17, 2026
Search Tags:#rag-security#prompt-injection#llm-safety#red-teaming#pii-protection#ai-guardrails#tool-security
One-click install
npx skills add https://github.com/roanbrasil/engineer-grade-agent-skills --skill ai-safety-guardrails
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ai-safety-guardrails
Source: https://github.com/roanbrasil/engineer-grade-agent-skills/tree/main/skills/ai-safety-guardrails
Command: npx skills add https://github.com/roanbrasil/engineer-grade-agent-skills --skill ai-safety-guardrails

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill solves critical security and safety gaps in production LLM applications, protecting against prompt injection, sensitive data exfiltration, harmful content generation, and insecure tool execution that can lead to data breaches, compliance violations, and system compromise.

Core Features & Use Cases

  • Multi-Layer Input/Output Guardrails: Enforce input length limits, detect known jailbreak patterns, run content moderation on user inputs, and validate LLM outputs against schemas, check for hallucinations, and moderate harmful content before delivery to users.
  • Prompt Injection Defense: Use structural patterns like delimiter isolation and sandwich defense to prevent user input from hijacking system instructions, with specialized protection for RAG pipelines where retrieved documents may carry adversarial content.
  • Tool and Data Safety: Implement least-privilege tool access, parameter validation to block path traversal and SQL injection, confirmation gates for irreversible actions, PII anonymization via Presidio, and full audit logging for all tool calls.
  • Red-Teaming and Monitoring: Integrate with open-source safety classifiers like LlamaGuard and NeMo Guardrails, run automated red-team test suites in CI/CD, and set up sanitized logging for anomaly detection.
  • Use Case Example: Use this Skill when building a customer support LLM chatbot that accesses internal order databases, to ensure malicious user inputs cannot inject commands to delete customer records or steal sensitive order data.

Quick Start

Use the ai-safety-guardrails skill to implement input validation, prompt injection defense, and output moderation for your new customer support LLM chatbot that accesses internal order databases.

Frequently Asked Questions about ai-safety-guardrails

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

FAQPage Schema
How do I prevent prompt injection in LLM chatbots that access internal databases?▼

To prevent prompt injection in LLM chatbots, apply structural defense patterns like delimiter isolation and sandwich defense to stop user input from hijacking system instructions. This ensures malicious queries cannot manipulate your database-accessing chatbot into executing unauthorized commands.

What is the best way to protect sensitive PII in production RAG applications?▼

The best way to protect sensitive PII in production RAG applications is implementing PII anonymization via Presidio alongside multi-layer input validation. This prevents sensitive data exfiltration through retrieved documents while maintaining compliance with data protection standards.

Can I use LlamaGuard and NeMo Guardrails for automated red-teaming in CI/CD?▼

Yes, you can integrate open-source safety classifiers like LlamaGuard and NeMo Guardrails to run automated red-team test suites in CI/CD pipelines. This enables continuous security testing to detect jailbreak patterns and harmful content generation vulnerabilities.

How do I secure external tool execution in agentic LLM workflows?▼

To secure external tool execution in agentic LLM workflows, implement least-privilege tool access controls and parameter validation to block path traversal and SQL injection. Add confirmation gates for irreversible actions and full audit logging for all tool calls.

Does multi-layer input validation work for preventing harmful content generation in customer-facing chatbots?▼

Multi-layer input validation works for preventing harmful content generation by enforcing input length limits, detecting known jailbreak patterns, and running content moderation on user inputs. Output validation also moderates harmful content before delivery to chatbot users.

Why do I need delimiter isolation for RAG pipelines when retrieved documents carry adversarial content?▼

Delimiter isolation is needed for RAG pipelines because retrieved documents may carry adversarial content that can hijack system instructions. Structural prompt defense patterns isolate external content from system prompts to prevent data exfiltration and unauthorized command execution.

Related Skills

Explore More Software Engineering Skills→
HechaoYannetHechaoYannet

create-architecture

Generate a master architecture document from GDDs, ADRs, and governance constraints.

Community
Advanced
SCStelzSCStelz
231

user-investigation

Analyze Entra ID user accounts for security issues and generate HTML reports.

Community
Advanced
ovachieverovachiever
52

nemo-guardrails

Enforce runtime safety rails for LLM applications using Colang 2.0 DSL.

Community
Advanced
moonmingmoonming
1

a6-plugin-ai-content-moderation

Configures APISIX AI content moderation plugins for LLM request and response filtering.

Community
Intermediate