ClawHub
Back to skill

Security audit

logic-pacer

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed prose-rewriting helper with a local read-only checker and no evidence of hidden data access, persistence, exfiltration, or destructive behavior.

Install this if you are comfortable with a prose-rewriting skill that may read local source/rewrite files for verification, run its bundled Python checker, and spawn a scoped blind-review subagent. Use it on intended prose snippets or files, and review the produced rewrite and flag block before accepting edits.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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 (2)

Context-Inappropriate Capability

Medium
Confidence
89% confidence
Finding
Granting Bash to a prose-rewriting skill unnecessarily expands its attack surface. Because the skill text explicitly instructs execution of a local command, adversarial or malformed inputs could influence file paths, subprocess usage, or side effects in ways that exceed the user's expectation for a text-editing skill.

Missing User Warnings

Medium
Confidence
87% confidence
Finding
The skill instructs execution of a verification script and has Write/Edit/Bash privileges, yet it does not clearly warn users about subprocess execution, file reads, or possible modification side effects. In a security context, hidden operational effects matter because users may provide sensitive paths or run the skill assuming it is a pure prose transformation.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.