skills.rest
maruakshay
maruakshayCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

rag-security

Audit RAG systems for ingestion poisoning, retrieval boundary failures, and metadata weaknesses.

4|Updated Apr 27, 2026
Search Tags:#security#rag#retrieval-augmented-generation#citation-verification#data-leakage#cross-tenant
One-click install
npx skills add https://github.com/maruakshay/mii-ai-security --skill rag-security
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: rag-security
Source: https://github.com/maruakshay/mii-ai-security/tree/main/skills/rag-security
Command: npx skills add https://github.com/maruakshay/mii-ai-security --skill rag-security

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

In a retrieval augmented generation (RAG) system, untrusted documents and weak metadata handling create surfaces for ingestion poisoning, cross-tenant leakage, and hallucinated grounding that misleads users.

Core Features & Use Cases

  • Detect and mitigate ingestion poisoning, ensure chunk provenance, and validate metadata to keep retrieved context trustworthy.
  • Enforce retrieval boundaries with server-side filters, provenance persistence, and robust citation verification across multi-tenant deployments.
  • Provide a repeatable review workflow with guardrails, checklists, and recommended quick wins to harden RAG pipelines in codebases and design patterns.

Quick Start

Apply the RAG security review to your codebase to identify ingestion poisoning, boundary violations, and metadata handling weaknesses.

Frequently Asked Questions about rag-security

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

FAQPage Schema
How do I prevent cross-tenant data leakage in a multi-tenant RAG system?▼

Prevent cross-tenant data leakage in multi-tenant RAG systems by enforcing server-side retrieval boundary filters and provenance persistence. This stops unauthorized context retrieval across tenants and keeps document trust intact.

What is ingestion poisoning in retrieval augmented generation pipelines?▼

Ingestion poisoning in retrieval augmented generation occurs when untrusted documents introduce malicious or misleading content into the knowledge base. Auditing chunk provenance and validating metadata keeps retrieved context trustworthy and prevents hallucinated grounding.

How do I verify citations in RAG outputs to prevent hallucinated grounding?▼

Verify RAG citations by applying structured citation verification and provenance tracking. This enforces document trust and ensures generated responses accurately reference their original retrieved chunks.

Can I audit my existing RAG codebase for retrieval boundary failures?▼

Yes, you can audit existing RAG codebases for retrieval boundary failures and metadata handling weaknesses. A repeatable review workflow identifies ingestion poisoning risks and provides guardrails, checklists, and quick wins to harden pipelines.

What's the best way to secure document metadata handling in cross-context RAG deployments?▼

Secure document metadata handling in cross-context RAG deployments by enforcing server-side controls and provenance persistence. This prevents stale-context exposure and document trust confusion by validating metadata throughout the retrieval pipeline.

Related Skills

Explore More Software Engineering Skills→
camelranchentertainmentcamelranchentertainment
1

rag-architect

Design production-grade retrieval augmented generation systems with chunking, embeddings, and vector stores.

Community
Advanced
agentsopeagentsope
287

agentsop-multi-tenant-rag

Enforce tenant isolation at vector store query boundaries in RAG pipelines.

Community
Advanced
AgentientAgentient
2

advanced-query

Design RAG queries with mandatory citation attribution and chunk-level provenance.

Official
Basic
pphatdevpphatdev

data-safety-and-privacy

Enforce data safety policies to prevent PII leakage and encrypt data at rest.

Community
Advanced