ClawHub

Skill Cortex Pub

Security checks across malware telemetry and agentic risk

Overview

This skill is transparent about its purpose, but it gives the agent broad power to find, install, run, remember, and remove other skills at runtime.

Install only if you are comfortable letting the agent manage other skills at runtime. Keep reflex behavior disabled or require explicit approval for every install and execution, avoid unreviewed GitHub candidates unless you inspect them first, and periodically review or delete the local cortex memory file.

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 (2)

Vague Triggers

Medium
Confidence
90% confidence
Finding
The trigger condition is broadly defined as activating whenever installed skills cannot complete the task, while also telling the agent to decide for itself whether it can 'handle it yourself.' That ambiguity can cause the skill to invoke package search and installation workflows in cases where simpler or safer handling was possible, increasing the chance of unnecessary third-party code acquisition.

Vague Triggers

Medium
Confidence
93% confidence
Finding
The semantic matching logic explicitly instructs the agent to use its own judgment rather than exact matching, then proceed directly into candidate selection and potentially reflex execution paths. In a skill whose purpose is to discover, install, and execute external skills from ClawHub or GitHub, this fuzzy routing increases the risk of misclassification, causing unintended invocation of untrusted capabilities or bypassing the user's expected review path.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal