Security audit
Security checks for vulnerabilities and agentic risk
The skill is a mostly transparent Agent personality/configuration tool, but it tells the agent to automatically modify a persistent SOUL.md configuration file without clear confirmation, scope, or rollback controls.
Review this skill before installing. It appears to be a configuration/personality diagnostic helper rather than malware, but do not let it modify SOUL.md unless you first see the exact diff and intentionally approve the change. Prefer a version that generates recommendations or patches for review instead of applying them automatically.
The final stage explicitly says it will compare results and '自动修改 SOUL.md 配置', confirming that the skill is designed to alter a sensitive configuration artifact automatically. In this context, the danger is elevated because the skill is framed as a personality diagnostic tool, so users may not expect it to persistently change system behavior or policy-like configuration as a side effect.
The skill description uses broad trigger phrases such as '测试 Agent 人格', '诊断 Agent 配置', and '优化 Agent 行为', which can overlap with ordinary user requests and cause the skill to activate in situations where the user did not explicitly request this workflow. Because the skill culminates in configuration changes, over-broad activation materially increases the chance of unintended analysis or modification of agent settings.
The core workflow states that Stage 5 outputs 'SOUL.md patches' and describes '自动修改' without any mention of user review, confirmation, rollback, or scope limits. A skill that can alter core configuration files without an explicit consent boundary creates a direct integrity risk and can silently change agent behavior in ways the user did not intend.
The usage instructions tell users to run the full diagnosis script but omit that the process may modify configuration files as part of Stage 5. This missing warning undermines informed consent and makes accidental destructive or behavior-altering changes more likely, especially when users treat the command as a harmless diagnostic.
The manifest describes a system that diagnoses an Agent's MBTI type and automatically generates configuration optimization suggestions. In this file, every branch merely echoes stage descriptions; notably Stage 5 advertises '分析对比 + 自动修改' and outputs 'SOUL.md patches', but no comparison, patch generation, or file modification logic exists.
This shell script is a code file, so safety-critical operations should include some form of user disclosure. Lines L29-L31 state that Stage 5 performs '自动修改' and outputs 'SOUL.md patches', but the script provides no confirmation prompt or explicit warning that user files may be changed.
Natural-language policy checks apply to all file types, including shell scripts with user-facing strings. The script presents all prompts and help text in Chinese only, with no opt-in, fallback, or documentation that this is intentionally limited to a Chinese-speaking context.
No suspicious patterns detected.
