Security audit

DNS查询免费版

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed DNS lookup helper, but its documentation is sloppy about output formats, activation scope, and DNS privacy implications.

Install only if you want an agent to run DNS diagnostic commands such as dig. Treat the JSON/structured-output claims as unreliable for the free edition, and avoid querying sensitive internal hostnames or investigation targets against public DNS resolvers unless you are comfortable disclosing those names to that resolver.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (6)

Description-Behavior Mismatch

Medium

Confidence: 93% confidence
Finding: The manifest promises structured/document-style output while the free-edition limitations later state that JSON structured output is unavailable. This inconsistency can mislead an agent or user into relying on unsupported output semantics, causing unsafe automation assumptions, parsing failures, or incorrect downstream handling of command results.

Description-Behavior Mismatch

Medium

Confidence: 78% confidence
Finding: The capability sections broadly claim support and structured handling for record/query operations, while later sections disclaim important free-edition functionality. Such contradictions can cause agents to invoke unsupported modes or expect controls that do not exist, leading to unreliable execution paths and possible misuse of exec-enabled tooling.

Intent-Code Divergence

Medium

Confidence: 95% confidence
Finding: The documented JSON output format directly contradicts the later statement that the free edition does not provide JSON structured output. This is dangerous because agents may parse non-existent structured output as trustworthy machine-readable data, producing logic errors, silent failures, or incorrect security decisions in automation workflows.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The activation text is overly broad and mixed with unrelated programming-assistant trigger language, making the skill easier to invoke outside its intended DNS-diagnostic scope. In an exec-capable skill, vague trigger conditions increase the chance of accidental activation and command execution on inputs that were not meant to be treated as DNS lookups.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The capability-scope keywords are excessively broad, ambiguous, and include unrelated terms, which can cause over-triggering of the skill. Because this skill advertises exec access, broad matching materially raises the risk of unintended command invocation and misuse in contexts unrelated to DNS diagnostics.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill omits a privacy/security warning that DNS lookups disclose queried domains or IP-derived PTR targets to external resolvers, especially when using third-party DNS servers. In an operational environment, this can expose internal hostnames, investigation targets, customer domains, or other sensitive metadata to external infrastructure.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.