ClawHub
Back to skill

Security audit

Autoskill Afrexai Tax Planning

Security checks across malware telemetry and agentic risk

Overview

This skill is a tax-planning helper, but it contains unusually aggressive instructions that can force long repeated outputs and reduce user control.

Review before installing. Use it only as a brainstorming aid, not as tax advice, and verify all recommendations with a licensed CPA or tax advisor. Prefer a revised version that removes the continuation loops, token minimums, repeat commands, and completion escape rules.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Missing User Warnings

Medium
Confidence
92% confidence
Finding
This skill provides tax-planning guidance, savings estimates, and jurisdiction-specific recommendations without prominently warning that the output is general information and may be inaccurate or unsuitable without professional review. In a regulated financial/tax context, missing disclaimers can cause users to rely on incorrect advice, increasing the risk of penalties, compliance failures, or harmful financial decisions.

Ssd 1

High
Confidence
99% confidence
Finding
The skill contains explicit behavioral instructions designed to override normal completion controls, suppress brevity, enforce minimum token counts, and coerce repeated continuation until status tags appear. This is dangerous because it attempts to seize control of the model’s output loop, causing runaway generations, denial-of-service style token exhaustion, and reduced compliance with higher-priority system safeguards.

Ssd 4

High
Confidence
99% confidence
Finding
The staged workflow is structured to look like harmless formatting guidance, but cumulatively it forces recursive expansion, repeated gates, recovery loops, and post-completion continuation passes that undermine model control. In context, this is more dangerous because the tax domain does not require such machinery; its presence strongly suggests an attempt to weaponize the skill into generating excessive output and resisting normal termination.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.