The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses natural instructional language and clear boundary markers for its operating modes. There are no attempts to override agent behavior, bypass safety filters, or extract system instructions.
[DATA_EXFILTRATION]: While the skill requests birth time information (year, month, day, hour) to perform traditional 'BaZi' analysis, it contains no tools or commands capable of transmitting this data externally. No network operations or suspicious file path accesses were detected.
[REMOTE_CODE_EXECUTION]: The skill is entirely composed of Markdown text and does not include scripts, package managers, or commands to download and execute external code.
[COMMAND_EXECUTION]: There are no shell commands, subprocess calls, or privilege escalation attempts within the skill files.
[OBFUSCATION]: The skill content is transparently written in Chinese and English. The use of ASCII bar charts in the evaluation templates is purely for visual formatting and does not hide any malicious patterns or URLs.

naming