RAG Engineering

Security checks across malware telemetry and agentic risk

Overview

This is an instructional RAG engineering guide with no executable code or hidden access requests; the main caution is how users handle documents they choose to index.

Safe to install as a methodology guide. When applying it to build a real RAG system, decide which document sources are allowed, avoid indexing secrets or unnecessary PII, enforce document-level permissions, and set retention rules for embeddings, metadata, logs, and cached responses.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (1)

Vague Triggers

Medium

Confidence: 93% confidence
Finding: The skill defines broad natural-language trigger phrases such as common help-style requests, which can cause the skill to activate on generic user queries that are not clearly asking for this specific capability. In an agent ecosystem, unintended invocation can lead to inappropriate routing, disclosure of internal methodology, or the skill taking over interactions where another safer or more specific skill should respond.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal