ClawHub
Back to skill

Security audit

sql-free

Security checks across malware telemetry and agentic risk

Overview

This SQL assistant is not clearly malicious, but it asks for execution and write-capable tooling while encouraging database design, DDL migration, and optional execution without strong safety boundaries.

Review this skill carefully before installing. Use it for drafting, explaining, or optimizing SQL, but do not let it automatically run SQL or migration scripts against production data. Treat schema and migration output as advisory only, require human review, and use read-only or disposable test databases where possible.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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
Findings (4)

Description-Behavior Mismatch

High
Confidence
94% confidence
Finding
The manifest explicitly says the skill is not suitable for database architecture design decisions, but the body advertises schema design and migration design as core capabilities. This scope contradiction can cause an agent or user to invoke the skill for higher-risk design tasks outside its declared bounds, increasing the chance of unsafe or low-governance database changes.

Intent-Code Divergence

High
Confidence
92% confidence
Finding
The internal documentation presents schema design as a primary capability even though the manifest excludes architecture/design decisions. In a skill with read/exec/write tools, this mismatch is dangerous because it can normalize use of the skill for sensitive database design or migration generation tasks without an accurate trust boundary.

Vague Triggers

Medium
Confidence
81% confidence
Finding
The activation description is broad and ambiguous, covering generic database operations, SQL queries, and storage management without clearly constraining what actions may be taken. In this skill context, broad triggering can cause accidental invocation in sensitive environments and expand from advisory SQL help into execution-adjacent behavior.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The usage flow includes optional execution/validation, but there is no prominent warning that execution may run SQL or commands with side effects. Because the skill declares exec/write capabilities and discusses migration and database operations, users or agents may proceed to execution without understanding risks such as destructive queries, data loss, or unintended access to live systems.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.