Security audit
Security checks for vulnerabilities and agentic risk
This Docker skill is mostly purpose-aligned, but it grants command and write authority while describing broad and partly unrelated automation without clear user control boundaries.
Install only if you intend to let the agent perform Docker administration. Review proposed commands before execution, especially builds, deploys, network changes, volume changes, deletes, and any operation involving credentials or production services.
The invocation guidance is overly broad, recommending use for code generation, programming assistance, debugging, testing, and deployment in general. Because the skill has exec/write tools, vague routing criteria can cause it to be selected in many unrelated conversations, increasing exposure to prompt injection, unsafe command execution, and unintended infrastructure modifications.
The skill is presented as Docker-focused, but these sections advertise unrelated capabilities such as code static analysis, dependency vulnerability checks, batch code review, and CI/CD integration without clear implementation boundaries. This kind of scope expansion can cause an agent to route broad development or security-analysis tasks to a skill that also has exec/write permissions, increasing the chance of unsafe command execution or misuse beyond the user’s intent.
The quick-start workflow tells users to invoke the skill and process results, but does not clearly warn that the skill can execute system commands and write changes affecting containers, images, networks, volumes, and deployment resources. In the context of a Docker skill, this omission is more dangerous because those actions can directly alter running services and host-connected assets, especially when paired with exec/write permissions.
The '主要功能' section broadens the skill from Docker operations into generic file processing and external API integration, which is inconsistent with a Docker management skill. In an agent ecosystem, this can mislead orchestration logic and users into granting or invoking a tool with powerful read/write/exec capabilities for tasks outside its intended domain, raising the risk of unintended filesystem changes or external data exfiltration.
No suspicious patterns detected.
