ClawHub

RAGLite

Security checks across malware telemetry and agentic risk

Overview

RAGLite appears to be a coherent RAG tool, but it needs review because it installs executable code from a mutable GitHub branch and can process sensitive documents through configured services.

Review the upstream GitHub repository before installing, and prefer a pinned commit or released package if available. Only run it on folders you are comfortable turning into a durable searchable cache, and confirm whether your OpenClaw gateway and Chroma server are local or remote before using it with sensitive documents.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding
The skill advertises a local RAG workflow, but the provided instructions explicitly install and rely on code fetched from a remote GitHub repository at runtime rather than implementing or vendoring the behavior locally. That creates a supply-chain and trust-boundary issue: users may believe they are running a transparent local-only skill, while in practice they execute unreviewed upstream code and an external engine path not clearly disclosed as core behavior.

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The install script pulls and installs code directly from a live GitHub repository branch (`@main`) at install time, which makes the installed code non-reproducible and exposes users to supply-chain compromise if the repository or branch is modified. The skill is described as local-first, so reaching out to a remote source during installation increases risk and undermines user expectations about locality and trust boundaries.

Context-Inappropriate Capability

Medium
Confidence
98% confidence
Finding
Installing executable Python code directly from a remote Git repository branch allows arbitrary upstream changes to be executed in the user's environment without prior review. Because `main` is mutable, an attacker who compromises the repo, maintainer account, or dependency path could deliver malicious code to every future installation.

VirusTotal

52/52 vendors flagged this skill as clean.

View on VirusTotal