Security audit
Security checks for vulnerabilities and agentic risk
This skill transparently generates a Markdown digest by reading top-level .md files, with no evidence of network access, credential use, persistence, or destructive behavior.
Install only if you want a Chinese-language digest tool that reads Markdown files in the current or specified directory. Avoid running it from a workspace containing private Markdown documents you do not want summarized in the report.
The skill instructs the agent to scan files in the current or specified directory, which is a file-reading capability, but it does not declare any explicit tool scope or permission boundaries. That mismatch can cause the skill to access more markdown files than a user expects, especially in shared or sensitive working directories, and makes review and enforcement of least privilege harder.
The trigger phrases are broad enough to overlap with ordinary summarization requests, so the skill may activate when a user only wants a conversational summary rather than a filesystem scan. That can unexpectedly enumerate local documents and expose filenames, headings, and derived metadata from unrelated markdown files.
The skill description does not warn users that it scans all markdown files in a directory and outputs file names, headings, and content-derived metadata. Without that disclosure, users may invoke it without understanding that local project structure or sensitive document metadata could be exposed in the generated report.
The activation examples do not distinguish between safe summarization tasks and requests that should initiate filesystem scanning. In practice, this ambiguity increases the chance of unintended skill execution and disclosure of document names, headings, and statistics from the working directory.
This Python file contains user-facing strings entirely in Chinese for the generated digest report and no option to select another language. The policy allows locale constraints only when they are user-selectable or clearly justified; neither is present here.
The name and examples are primarily in Chinese, and the invocation phrases and descriptive text indicate a Chinese-default experience except for a single English trigger. There is no statement that users may choose their preferred language or that the locale limitation is intentional and justified.
No suspicious patterns detected.
