humanizer
Pass
Audited by Gen Agent Trust Hub on May 10, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists exclusively of markdown documentation and logic rules for text transformation. No executable scripts (.sh, .py, .js) or binaries are included.
- [DATA_EXFILTRATION]: Analysis found no network operations (curl, wget, fetch) or access to sensitive file paths like credentials or environment variables.
- [PROMPT_INJECTION]: The skill processes external text input via the
/humanizeand/人话commands. Although it lacks specific delimiters to isolate user input from instructions, the absence of any functional tools (e.g., filesystem or network access) prevents potential indirect prompt injections from escalating into system-level threats. - [REMOTE_CODE_EXECUTION]: There are no patterns suggesting the download or execution of remote code, and no external package managers are invoked.
Audit Metadata