Security audit

容器检查专业版

Security checks across malware telemetry and agentic risk

Overview

This skill is a container operations helper, but it asks for broad command and remote-management authority with unclear limits around SSH credentials, bulk host actions, reports, and scheduled data retention.

Review before installing. Use this only in environments where the agent is allowed to inspect containers and run remote SSH commands. Prefer non-root SSH accounts, explicit host allowlists, strict SSH host-key checking, protected config files, and controlled report locations. Do not use --hosts all, remote exec, scheduled audits, or long retention unless those behaviors are intended and authorized.

SkillSpector

Vulnerability Patterns

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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Intent-Code Divergence

Medium

Confidence: 94% confidence
Finding: The security note says API keys should be provided via environment variables and not hardcoded, but the documented model elsewhere says the PRO version does not use external API keys and instead stores database credentials in a local configuration file. Conflicting guidance causes operators to place secrets in inconsistent locations, increasing the chance of accidental exposure or insecure deployment practices.

Intent-Code Divergence

Medium

Confidence: 92% confidence
Finding: The skill documents remote management over SSH, but the network-security section generically instructs use of HTTPS and SSL certificate validation. That mismatch can mislead users about the real trust boundary and transport protections, causing them to overlook SSH host-key verification and other SSH-specific safeguards.

Vague Triggers

High

Confidence: 96% confidence
Finding: The description contains broad activation phrases about data analysis, reporting, visualization, compliance, encryption, and general workflow use that go well beyond container diagnostics. In an agent environment, this can cause unintended invocation in unrelated contexts, granting a skill with exec/write/remote-management capabilities access when the user did not specifically request container operations.

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The capability coverage section lists vague tokens like 'Use' and 'when' as scenario keywords, making routing criteria effectively ambiguous. This broadens activation scope and can cause the skill to be selected for unrelated prompts, which is more dangerous here because the skill advertises exec, write, and remote-management capabilities.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The documentation promotes remote command execution and report/file output but does not clearly warn about execution scope, target authorization, possible side effects, or data sensitivity of generated artifacts. Users may trigger actions against multiple hosts or produce files containing sensitive operational details without informed consent, especially in an agentic setting.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.