芒格.skill - 查理·芒格思维操作系统

This skill is coherent with its stated purpose (a Charlie Munger-style thinking advisor) and is low-risk from a system/credential perspective because it is instruction-only and asks for no secrets. However: - It mixes a clear disclaimer ('this is not Munger') with an identity card that reads 'I am Charlie Munger' and instructs the agent to respond in the first person. That inconsistency increases the chance people will treat outputs as if they were the real person's views. Consider whether you need that level of impersonation or if you prefer a clear, persistent disclaimer on each session. - The skill mandates a blunt, insulting style and to avoid meta-analysis unless explicitly asked. That can produce terse, potentially offensive, or legally/problematic statements (for example, harsh judgments about individuals or policies). Avoid using it in contexts that require empathy, diplomatic wording, or when you need careful, sourced factual claims (investment/legal/medical decisions). - The one-time-disclaimer rule is risky for multi-turn conversations or when outputs are exported/shared later. Ask the publisher to: (a) remove the contradictory 'I am Charlie Munger' phrasing or make the persona explicit but clearly fictional; (b) require the disclaimer appear at each new session or when content is shared; (c) soften or allow configurable tone settings for sensitive contexts. If you proceed, treat any prescriptive advice from the skill as opinionated roleplay, verify factual claims independently, and avoid giving it access to sensitive data or using it for legally/financially binding recommendations.