ClawHub
Back to skill

Security audit

find-skills-combo

Security checks across malware telemetry and agentic risk

Overview

This skill is a coherent skill-discovery helper, but it broadly routes user requests into external skill search and presents global no-confirm install commands as normal next steps.

Review candidate skills and their sources before installing anything. Prefer install commands without `-g -y`, require explicit approval for each install or update, and treat global skill installs as persistent changes to the agent environment.

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 (4)

Vague Triggers

Medium
Confidence
88% confidence
Finding
The metadata says to use this skill whenever the user asks broad questions like 'how do I do X' or 'find me a skill', and even claims it supersedes simpler search. That can cause the agent to invoke a more complex workflow than necessary, increasing the chance of irrelevant searches, unwanted recommendations, or steering the user into installs they did not explicitly request.

Vague Triggers

Medium
Confidence
81% confidence
Finding
The 'when to use' criteria overlap heavily with ordinary help requests and only weakly constrain when fallback should be used. In a skill that can search for and recommend installations, ambiguous invocation increases the risk of unnecessary tool use and overbroad capability discovery on simple prompts.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The documentation normalizes `npx skills add <package> -g -y`, which performs global installation and suppresses confirmation. Presenting auto-install syntax without a safety warning makes it easier for an agent or user to execute package installation non-interactively, increasing supply-chain and unwanted system modification risk.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The recommended final output explicitly tells the agent to present global, confirmation-skipping install commands as the normal next step for both strategies. In context, this is more dangerous because the skill is designed to discover and recommend third-party skills, so it directly couples discovery with low-friction installation.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.