ClawHub
Back to skill

Security audit

ClawGuard

Security checks across malware telemetry and agentic risk

Overview

ClawGuard appears to be a local, read-only skill scanner, with the main caveat that its documentation is broader than the checks actually implemented.

This is reasonable to install as a lightweight local scanner. Run it only on skill folders you intend to inspect, verify the actual scan.py path after installation, and do not treat a PASS result as proof that another skill is safe; manually review high-impact skills and any findings before installing them.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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)

Lp3

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding
The skill declares no required permissions or environment access, yet its documented operation and direct CLI usage imply local file reads and shell execution via `python3 .../scan.py <path>`. This mismatch can mislead users and any permission-gating system about the skill's real capabilities, reducing transparency and weakening install-time trust decisions.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The natural-language invocation examples use broad phrases like 'check', 'audit', 'scan', and 'inspect a skill before installing' that can plausibly appear in ordinary conversation. In agent systems with fuzzy matching, this increases the chance of unintended or surprise activation, causing the skill to read local directories or run analysis when the user did not explicitly mean to invoke it.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The metadata description advertises several generic trigger phrases, including 'is this skill safe' and 'check before installing', without strong qualifiers. Because these phrases are broad and common, they may overlap with routine discussion and increase accidental activation risk, especially in ecosystems where skills are selected from descriptive text.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.