The Agent Skills Directory

[PROMPT_INJECTION]: The skill provides instructions for the agent to adopt a specific writing persona. There are no attempts to bypass safety filters, override system constraints, or extract internal prompt data.
[DATA_EXFILTRATION]: No network operations, sensitive file path access, or credential harvesting patterns were found in the skill files.
[REMOTE_CODE_EXECUTION]: The skill does not contain commands that download or execute external code at runtime.
[COMMAND_EXECUTION]: No dangerous shell commands or subprocess calls are present. The only shell command mentioned is a setup instruction for the platform in the documentation.
[CREDENTIALS_UNSAFE]: No hardcoded API keys, tokens, or passwords were found.
[OBFUSCATION]: The content is presented in clear text (Chinese and English) without the use of Base64 encoding, zero-width characters, or other hidden markers.
[DYNAMIC_CONTEXT_INJECTION]: The skill main file does not use any dynamic command execution syntax (such as !command).

hemingway-perspective