Wecomcli Contact Skill Security Audit

The visible command only single-quotes the placeholder `<JSON 参数>`; it does not show an implementation that directly concatenates a user-supplied keyword into a shell command. If an executor replaces that placeholder with an unescaped string, a quote in the keyword could enable shell injection. However, the command may instead be run using safe JSON serialization and a structured argument array. Without the script or command-building code, the risk cannot be confirmed. Users can ask for evidence that arguments are passed as structured argv values with correct JSON and shell-boundary encoding.

The skill queries and returns directory data including unique user IDs, names, email addresses, positions, and department paths, and enables a full-match mode when the requester explicitly asks for a complete list. The visible rules do not explain how the requester's authorization to access this personal and organizational data is verified. They refer to a separate prerequisite skill, but its contents are not provided. If the underlying account has broad access or these fields are shown directly, sensitive directory data could be disclosed. Users can ask whether the shared checks enforce identity, authorization, and minimum-field output.