Security audit
Security checks for vulnerabilities and agentic risk
This appears to be a real Markdown link checker, but its security metadata understates the file, execution, and optional network behavior it actually contains.
Review this skill before installing because its declared permissions do not match the script and rationale. It should be limited to documentation directories, and external link checking should only be used intentionally with --net. Ask the publisher to align permissions, remove unnecessary Write authority, and clarify whether network checking is supported.
The manifest describes an offline Markdown link auditor with disabled network/filesystem/subprocess permissions, but the risk rationale claims API usage, file access, and local execution. This inconsistency can mislead reviewers and automated policy systems about the true behavior or expected trust boundary, making unsafe approval or improper deployment controls more likely.
The risk rationale explicitly states capabilities such as API calls, file reads/writes, and local execution, while the manifest declares network, filesystem, and subprocess permissions as false. Contradictory security declarations undermine policy enforcement, operator trust, and auditability, and may allow the skill to be handled as lower risk than its documentation suggests.
The file contains user-facing descriptions, help text, status messages, and error output entirely in Chinese, including the CLI description and runtime prints. For a general-purpose Markdown link auditing script, this imposes a specific language/locale without any opt-in or documented region-specific justification, which matches the natural-language policy violation criteria.
No suspicious patterns detected.
