嘴替 Skill
AdvisoryAudited by Static analysis on Apr 30, 2026.
Overview
No suspicious patterns detected.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent may force clarification or prompt rewriting even when you asked for a direct answer or wanted to bypass the skill.
This makes the skill mandatory for all user questions and explicitly prevents skipping, which can override the user's immediate intent and normal agent routing.
所有用户提问必须经过本技能，不可跳过
Make activation opt-in or clearly conditional, and allow users to skip the preprocessor for direct, already-clear, or time-sensitive requests.
Installing it could permanently change how the agent handles future conversations by routing all questions through this preprocessor.
The skill asks to modify persistent agent-level files so its activation rules continue to affect future behavior beyond a single invocation.
安装后须同步更新 AGENTS.md / SOUL.md ... 写入嘴替激活规则
Do not edit persistent agent files automatically; require explicit user approval, show the exact diff, and provide a clear rollback step.
The skill may remember preferences or assumptions from past questions and reuse them later, which could be inaccurate or unwanted.
The skill describes optional user profiling based on historical query patterns and preferences, which may persist context across tasks.
本技能可记录每个用户的历史 query 模式
Use profile memory only with user consent, document where it is stored, allow review/deletion, and keep current user instructions higher priority than history.