Security audit
Security checks for vulnerabilities and agentic risk
This archiving skill is not clearly malicious, but it combines persistent full-content storage with broad read, browser, and command-execution authority that is not tightly scoped.
Review this skill before installing. Use it only for content you are allowed to copy and retain, avoid archiving sensitive or restricted material, and restrict command execution to explicit, fixed commands for archive setup or maintenance. Expect local persistent files under ~/archive and plan your own deletion, redaction, and backup policy.
The top-level description uses expansive, ambiguous conditions such as general AI model calling, intelligent dialogue, agent orchestration, and LLM application use. This weakens the principle of least privilege because a broadly described skill with powerful tools may be selected for many unrelated tasks, increasing exposure to unintended file, browser, or execution actions.
The skill declares a very broad set of activation keywords, including unrelated AI, workflow, and SEO-style scenarios, far beyond simple content archiving. Overbroad routing language can cause the skill to be invoked in inappropriate contexts, where its read/exec/browser capabilities may be granted under false expectations and used on tasks outside its safe design scope.
The description states "支持中文交互" and presents the skill as operating in Chinese, but does not offer a language choice or explain a justified locale restriction. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation.
The skill encourages capturing and permanently storing complete external content, but it does not adequately warn about persistence, privacy, copyright, or handling of sensitive material. In context, this is risky because an agent may archive personal data, proprietary content, or restricted material into long-lived local storage without informed consent or retention controls.
Earlier instructions state that archiving should extract full content and generate a 2-3 line summary as part of the standard workflow. Later, the '已知限制' section says the free version requires an upgrade for '自动摘要生成', which directly conflicts with the earlier documented behavior for this same free skill.
The manifest presents the skill as a narrowly scoped content-archiving tool, but later sections broaden it into generic automation, API integration, file handling, and command execution. This scope expansion can cause an agent or user to authorize actions far beyond the expected trust boundary, increasing the chance of unsafe execution or data access under a misleading description.
The skill explicitly advertises system command execution even though its stated purpose is content archiving. In an agent setting, unnecessary exec capability materially raises risk because user-provided content, URLs, or metadata may be transformed into shell actions, enabling command injection, filesystem abuse, or unintended host interaction.
