ClawHub
Back to skill

Security audit

Docker基础工具免费版

Security checks across malware telemetry and agentic risk

Overview

This Docker helper is intended for container work, but it can run commands that delete containers, images, and saved data without clear approval steps.

Review this skill before installing in any environment with important containers, images, or volumes. Use it only for explicit Docker tasks, and require the agent to list targets and ask before running removal, prune, down -v, or volume deletion commands.

SkillSpector

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

Vague Triggers

Medium
Confidence
82% confidence
Finding
The trigger condition is broad enough to activate this skill for common development and deployment requests, increasing the chance that an agent will invoke `exec` and run Docker commands in contexts where the user did not intend host-level container operations. Because the skill has execution capability and includes destructive lifecycle commands, overbroad routing materially raises the risk of unintended command execution and environment modification.

Missing User Warnings

High
Confidence
96% confidence
Finding
The skill documents destructive commands like force-removing containers and pruning stopped containers without prominent warnings, dry-run guidance, or confirmation requirements. In an agent with `exec` access, these examples can translate directly into irreversible deletion of running or stopped workloads, causing service disruption and loss of debugging state.

Missing User Warnings

High
Confidence
97% confidence
Finding
Image and cleanup commands such as `docker rmi`, `docker image prune`, and especially `docker image prune -a` can irreversibly remove local artifacts needed for development, recovery, or rollback. In the context of an executable agent skill, presenting these commands without safeguards materially increases the risk of unintended data loss and operational disruption.

Missing User Warnings

High
Confidence
98% confidence
Finding
`docker-compose down -v` deletes associated volumes, which may contain persisted databases and application state. Because the skill targets development workflows and is allowed to execute commands, omitting an explicit warning makes accidental permanent data loss significantly more likely.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.