Security audit

Docker V1基础工具免费版

Security checks across malware telemetry and agentic risk

Overview

This Docker helper is not malicious, but it gives broad command-execution guidance and includes destructive Docker actions without clear user safeguards.

Install only if you specifically need Docker V1 or docker-compose V1 help, and keep the agent limited to that scope. Require an explicit review before any remove, prune, down -v, push, rebuild, sudo, or login command; back up important volumes first; and do not paste registry passwords or tokens into commands or logs.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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

Findings (9)

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The skill advertises generic external API integration and network communication capabilities that are not necessary for a Docker V1 local-management skill. This scope expansion can cause an agent to route unrelated requests to a skill with exec privileges, increasing the chance of unintended network access or command execution beyond the user's Docker task.

Description-Behavior Mismatch

Medium

Confidence: 93% confidence
Finding: The documentation claims generic file processing and write-oriented behavior beyond the manifest's stated Docker V1 essentials scope. In a skill that has exec access, overstating file-handling powers can make the agent apply the skill to broader tasks and potentially modify files outside intended container-management workflows.

Description-Behavior Mismatch

Low

Confidence: 82% confidence
Finding: Marketing the skill as suitable for full-process automation and generic repeated-task execution broadens its apparent authority beyond Docker V1 essentials. This increases unsafe invocation risk because an agent may over-select the skill for unrelated automation requests while it still has command-execution capability.

Vague Triggers

High

Confidence: 97% confidence
Finding: The trigger condition is overly broad, covering general code generation, debugging, testing, and deployment scenarios rather than narrowly scoped Docker V1 maintenance tasks. Because the skill exposes exec, broad routing guidance can cause it to be invoked for common developer requests where shell execution and destructive Docker commands are unnecessary and risky.

Vague Triggers

High

Confidence: 97% confidence
Finding: The manifest description includes ambiguous guidance such as use for code generation, programming help, debugging, testing, and deployment, which is much broader than Docker V1 basics. In skill-selection systems, this can lead to over-invocation of an exec-capable skill for unrelated developer workflows, expanding the attack surface and chance of harmful command execution.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The skill documents force-removal and prune commands that delete containers without warning users about irreversibility, data loss, or service disruption. In an agent setting, examples often become executable suggestions, so lack of warnings and confirmation gates materially raises the risk of accidental destructive operations.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: Image deletion and prune commands are presented as routine actions without warning that they can irreversibly remove local images and affect rebuild time, rollback ability, or dependent workflows. In an exec-enabled skill, this omission can translate into unsafe automated cleanup behavior.

Missing User Warnings

High

Confidence: 98% confidence
Finding: The documented use of docker-compose down -v is especially dangerous because it removes persistent volumes and can permanently destroy application data. Presenting this without a clear data-loss warning or confirmation requirement is a significant safety issue in a container-management skill.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The registry login example encourages credential entry for private registries but does not warn about secure credential handling, shell history exposure, or the need for credential helpers. In an agent-assisted environment, this can lead users to paste secrets directly into commands or logs.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.