ClawHub
Back to skill

Security audit

Property Tax Appeal

Security checks across malware telemetry and agentic risk

Overview

This skill is a coherent property-tax appeal packet builder with disclosed local file generation, research, and validation behavior, though users should be careful with the optional overwrite flag and English/PDF limitations.

Install only if you are comfortable with a skill that researches public property/tax sources, processes sensitive property appeal facts, and creates local Markdown/PDF files. Use a fresh case directory, keep private documents separate, review every generated statement before filing, and avoid --force unless you have manually confirmed the existing outputs are for the same case.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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 (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding
The skill instructs the agent to read and write files, access the network, and invoke shell commands, but there is no declared permission model accompanying those capabilities. That mismatch creates a trust and governance gap: operators or users may not realize the skill can touch the filesystem, browse external sites, and execute local scripts, increasing the risk of unintended data exposure or unsafe execution if the environment does not separately sandbox those actions.

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The CLI help and surrounding error text promise that --force should only be used after verifying the existing output belongs to the same case, but the implementation does not actually perform any such check before overwriting destination files. In automation or multi-user workflows, this can lead to accidental destruction or replacement of unrelated Markdown/PDF outputs, undermining integrity and auditability of appeal packets.

Natural-Language Policy Violations

Medium
Confidence
79% confidence
Finding
The skill defaults to producing an English attachment without obtaining user preference or confirming that English is acceptable. This can create safety and fairness issues by causing inaccurate or inaccessible filings for users who need another language, and it may pressure users into a language choice that affects comprehension of legal or tax documents.

Natural-Language Policy Violations

Medium
Confidence
88% confidence
Finding
The PDF generation flow fails closed on unsupported characters and instructs the user to use an 'official English spelling' or skip PDF generation. That can coerce transliteration or anglicization of names and addresses in a legal context, creating integrity risk if the rendered filing no longer exactly matches official records or owner identity details.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.