ClawHub
Back to skill

Security audit

Score Analyzer

Security checks across malware telemetry and agentic risk

Overview

This skill is mostly aligned with score-report generation, but it asks agents to run missing local scripts and may install a system font package during use.

Review before installing. Use it only in a dedicated workspace, with student data you are authorized to process. Do not allow the apt install font step unless you intentionally want system fonts installed. Also verify the referenced scripts are trusted and actually come from the skill or your chosen workspace before letting the agent run python3 scripts/*.py commands.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The skill instructs the agent to run an external package-management command (`apt install fonts-noto-cjk`) even though the skill is supposed to perform local score analysis with Python tooling. Allowing environment-modifying package installs expands the trust boundary, can change the host system, and is unrelated to the core task of analyzing spreadsheet data.

Context-Inappropriate Capability

Medium
Confidence
96% confidence
Finding
The skill contains capability for system-level environment modification by instructing installation of OS packages during execution. This is unnecessary for score analysis/report generation and creates avoidable risk of privilege misuse, system drift, or abuse in environments where agents may execute shell commands automatically.

Missing User Warnings

Low
Confidence
78% confidence
Finding
The skill processes uploaded Excel files containing student scores and identifiers, but the description does not warn about handling potentially sensitive educational data. Without an explicit privacy notice, users may provide regulated or confidential records without understanding that the files will be read, transformed, and reported on.

Missing User Warnings

Low
Confidence
74% confidence
Finding
The instructions direct the creation and overwriting of multiple files under `reports/` without explicitly warning the user about filesystem changes. In agent-executed environments, this can surprise users, overwrite prior outputs, and leave persistent artifacts containing student data.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.