ClawHub
Back to skill

Security audit

Query Rewrite

Security checks across malware telemetry and agentic risk

Overview

This skill is a prompt-only query rewriting helper for RAG searches, with its automatic retrieval-time behavior disclosed and no evidence of hidden code, persistence, credential access, or destructive actions.

Before installing, understand that retrieval searches may be expanded with recent conversation context to improve recall. This is appropriate for a RAG helper, but avoid enabling it on workflows where recent chat context should never be sent into memory, wiki, or vector search systems.

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
86% confidence
Finding
The README states the skill activates automatically before any RAG retrieval call, which gives it broad interception scope over search inputs across workflows. Even though the described behavior is legitimate query rewriting, this increases attack surface and can cause unintended query transformation, context leakage from prior turns into searches, or retrieval manipulation if the rewrite logic is wrong or later modified unsafely.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.