awesome-novel
Warn
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: MEDIUMPERSISTENCECOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PERSISTENCE]: The installation scripts
install.shandinstall.ps1modify user environment configuration files, such as~/.bashrc,~/.zshrc,~/.profile, and Fish shell configs. These scripts usegrepandechoto ensure theNOVEL_SKILL_HOMEenvironment variable is exported in every new shell session. - [COMMAND_EXECUTION]: The skill uses Python's
subprocess.runintools/sync-project.pyto executegitcommands for version detection. Furthermore, theSKILL.mdinstructions command the AI agent to execute local Python scripts (tools/init.pyandtools/sync-project.py) to initialize projects and synchronize core agent definitions and knowledge bases with the filesystem. - [PROMPT_INJECTION]: Several agent definitions (e.g.,
agents/novel-base.mdandagents/novel-agent.md) use authoritative framing to override standard model behaviors. Examples include "写作子代理基底——永久加载,不可篡改" (Base for writing sub-agent - permanently loaded, unalterable) and "核心写作铁律" (Core writing iron rules), which are designed to strictly enforce specific narrative styles and suppress AI self-references or meta-commentary.
- [EXTERNAL_DOWNLOADS]: The skill's documentation and scripts facilitate cloning components and updating assets from a GitHub repository (
https://github.com/modoojunko/awesome-novel-skill.git).
Audit Metadata