The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions contain no attempts to bypass agent safety filters or override system behavior. Instead, it includes strong internal constraints () designed to ensure high-fidelity responses and prevent the agent from making unsupported claims.
[DATA_EXFILTRATION]: No unauthorized data access or exfiltration patterns were identified. The skill instructs the agent to provide helpful links to a specialized digital library (fojin.app), which is appropriate for its educational purpose and does not involve the transfer of sensitive user or system data.
[REMOTE_CODE_EXECUTION]: While the skill references optional helper scripts (cite.py, query.py), these are not provided in the package and their described use is for standard local CLI metadata retrieval. No remote script execution or unverifiable package installations are present.
[COMMAND_EXECUTION]: There are no shell commands, dynamic context injections (!command), or privilege escalation attempts within the provided files.
[SAFE]: The skill demonstrates best practices for AI agent extensions, including clear versioning, metadata documentation, and structured source material for grounded generation.

master-huineng