ClawHub
Back to skill

Security audit

Database Design Hardened

Security checks across malware telemetry and agentic risk

Overview

This is a local database-design helper whose scripts generate SQL/text output and keep a small local command history, with no evidence of hidden network access or destructive execution.

Install only if you are comfortable running local shell scripts from this publisher. Be aware that one script keeps a local history file of commands and first arguments; avoid passing secrets or sensitive schema details as arguments, and delete ~/.local/share/database-design/history.log if needed. Review generated SQL carefully before applying it to any real database, especially migrations, index drops, and schemas involving sensitive data.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Missing User Warnings

Medium
Confidence
98% confidence
Finding
The script logs command activity and user-supplied arguments to a persistent file under the user's data directory without any notice, consent, or sanitization. In a database-design context, arguments may include schema names, table names, file paths, sample data references, or potentially sensitive query fragments, creating an unnecessary privacy and confidentiality risk if the local machine is shared or the logs are later collected.

VirusTotal

67/67 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.