write-workflow

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill documents how to install required dependencies from external sources using a CLI tool. Specifically, it references https://github.com/blader/humanizer.git and https://github.com/op7418/Humanizer-zh.git. These are presented as standard external dependencies for natural language processing tasks and are handled via a package manager (npx skills add).
  • [COMMAND_EXECUTION]: The documentation provides examples of shell commands for dependency installation (e.g., npx skills add FuDesign2008/open-skills -g --skill <name> --yes). These are instructional examples for the user to set up their environment and are not automatically executed by the skill in a hidden or malicious manner.
  • [SAFE]: The skill follows best practices by implementing an explicit eight-stage workflow (clarify, analyze, explore, review, outline, execute, verify, retrospect) with clear 'hard pauses' for user approval, especially before critical writing steps. It explicitly prohibits 'silent degradation' and requires verification of dependencies at startup.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 07:39 AM
Security Audit — agent-trust-hub — write-workflow