The Agent Skills Directory

[SAFE]: The skill consists entirely of natural language instructions and structural frameworks for conversation coaching. There is no source code, shell commands, or script execution involved.
[DATA_EXFILTRATION]: No network operations (such as curl or wget) or external URL references were detected. The skill does not attempt to access or transmit any user data.
[CREDENTIALS_UNSAFE]: No hardcoded secrets, API keys, or sensitive environment variable accesses are present in the instructions.
[PROMPT_INJECTION]: The instructions include standard behavioral constraints (e.g., '禁止做的事' / 'Things forbidden') to guide the AI's persona and prevent it from giving long-winded or unhelpful advice. These are typical safety and quality alignment measures rather than malicious bypass attempts.
[EXTERNAL_DOWNLOADS]: The skill does not define any external dependencies, package installations (NPM/PyPI), or remote script fetches.
[COMMAND_EXECUTION]: There are no shell commands or subprocess calls defined in the documentation. References to other internal commands (like /taxue-calm) are navigational pointers to related skills within the same ecosystem.

taxue-speak