How does this skill prevent prompt injection?

It implements Layer 5 (LLM Orchestration) which flows identifiers around the LLM rather than through it, and Layer 6 (Output Validation) to verify all AI-generated content before use.

Is this framework based on industry standards?

Yes, the patterns are derived from NIST (Defense in Depth), Google BeyondCorp (Zero Trust), and AWS IAM (Least Privilege) best practices.

What are the 8 layers of defense in this architecture?

The framework covers Edge protection, Gateway/Auth, Input validation, Authorization (RBAC/ABAC), Data Access, LLM Orchestration, Output validation, and Observability.

Does it support multi-tenant applications?

Yes, it includes specific patterns for immutable request contexts and tenant-scoped repositories to ensure strict data isolation and prevent cross-tenant access.

AI Defense-in-Depth

Name: AI Defense-in-Depth
Author: yonatangross

byyonatangross

•

Security & Testing

Implements a robust 8-layer security architecture to harden AI pipelines and protect LLM integrations from end-to-end.

About

The Defense-in-Depth skill provides a comprehensive security framework for AI applications, moving beyond simple prompt engineering to secure every boundary from the network edge to final storage. By implementing eight distinct validation layers—including edge protection, immutable request contexts, strict tenant isolation, and output guardrails—this skill ensures that no single point of failure can compromise your system. It is an essential tool for developers building production-grade AI services that require rigorous PII protection, multi-tenant security, and resilient LLM orchestration patterns that prevent data leakage and prompt injection.

Key Features

29 GitHub stars
Standardized security implementation checklists
LLM orchestration patterns to prevent prompt injection
Immutable Request Context for secure identity flow
8-Layer Security Architecture (Edge to Observability)
Multi-tenant isolation and query-level filtering

Use Cases

Hardening production LLM applications against prompt injection and data leaks
Building audit-ready AI systems with sanitized logging and tracing
Implementing strict multi-tenant isolation in RAG-based AI services

About

Key Features

29 GitHub stars
Standardized security implementation checklists
LLM orchestration patterns to prevent prompt injection
Immutable Request Context for secure identity flow
8-Layer Security Architecture (Edge to Observability)
Multi-tenant isolation and query-level filtering

Use Cases

Hardening production LLM applications against prompt injection and data leaks
Building audit-ready AI systems with sanitized logging and tracing
Implementing strict multi-tenant isolation in RAG-based AI services