Security audit
Security checks for vulnerabilities and agentic risk
This skill has a coherent X-to-Obsidian daily report purpose, but it asks for broad execution/write authority and has unclear fallback storage and scope controls.
Review this before installing if you do not want an agent using an existing X login session, running shell commands, or writing reports outside a clearly named Obsidian folder. Use it only with explicit account/keyword scope and a confirmed output path.
The trigger description is overly broad and mixes X-content collection with unrelated SEO, keyword analysis, traffic optimization, and ranking-improvement use cases. In an agent environment, vague activation criteria can cause the skill to be invoked in contexts the user did not intend, leading to unnecessary tool use, data collection, or writes.
The description states "支持中文交互" as a fixed behavior, and the document does not offer any language or locale opt-in choice. This creates a natural-language policy concern because it appears to impose a specific language rather than letting the user choose.
The example trigger phrase '采集今日 AI 内容日报' is too underspecified for a skill that may read configuration, access external content, and write files. Without explicit scope constraints such as source limits, account lists, output path, and whether external access is allowed, the agent may over-collect or perform actions the user did not specifically authorize.
The documentation says the free version only writes to Obsidian, but later permits fallback writing to a workspace mirror directory when the vault is not writable. That creates an undeclared alternate data sink, which can expose collected content in unintended locations and weakens user control over where data is stored.
The skill advertises broad command-execution, file handling, and API-integration capabilities that exceed the narrowly stated task of collecting X content and writing an Obsidian report. In an agent setting, this kind of overbroad capability framing can normalize or justify risky tool use, increasing the chance that the agent executes unrelated shell commands or accesses resources beyond the user’s intent.
The X login handling section explicitly says the free version skips blocked sources instead of automatically waiting for login. Elsewhere, the document broadly claims '全流程自动', '内置错误恢复', and '自动化执行', which overstates the automation level and conflicts with the documented manual-login prerequisite for a core data source.
No suspicious patterns detected.
