Security audit
Security checks for vulnerabilities and agentic risk
The skill is not clearly malicious, but it needs review because it stores persistent user preferences while its trigger conditions and exec permission documentation are inconsistent.
Install only if you are comfortable with a skill that can write persistent preference files in your home directory and whose manifest grants exec despite saying exec is unnecessary. Review or narrow the trigger conditions and permissions before using it broadly.
The trigger condition is both mismatched and overly broad, making accidental invocation likely. In the context of a skill that can write local files and is declared with exec, erroneous triggering materially increases risk because the agent may persist data or perform actions during unrelated workflows.
The documentation explicitly claims the skill is pure Markdown and does not require exec, while the manifest still declares exec. This is a high-risk deception/misconfiguration pattern because reviewers and users may approve a supposedly non-executing skill that can in fact run shell commands.
该技能的 displayName、summary、description 以及主体说明均以中文固定呈现，且未说明是否支持按用户偏好切换界面/说明语言。对于面向“任意AI Agent”的通用技能，这种默认单一语言呈现可能构成语言/locale 约束而缺少用户选择。
The skill markets itself as local Markdown-only preference storage with no cloud/API dependency, yet the manifest grants exec and the body includes shell command examples. This mismatch expands the effective trust boundary: an agent may execute commands under a skill the user reasonably expects to be file-only, enabling unintended filesystem or process actions.
The feature sections say the skill supports export, save, convert, modify, reset, and import operations. Later, the known limitations for the free edition explicitly state that preference import/export and cross-Agent migration are not available. These statements conflict about what the free skill actually does.
The trigger conditions say to use the skill for design/visual-creation tasks, which is unrelated to communication-preference learning. Overbroad or mismatched triggers can cause the agent to invoke this skill in inappropriate contexts, increasing the chance of unintended local file writes or command use outside the user’s expectations.
The skill states it will automatically create a directory and write user preferences to local files, but it does not present a strong, explicit consent notice at the point of storage. This can lead to silent persistence of personal communication preferences, which are user data, without sufficiently informed approval.
