Grok Bot软件写代码交付AI 评分 · 85/100AI 辅助编辑·未发现明显提示词注入风险X

让 AI 用五档心智模型给你做使用方式绩效评估

复制提示词 → 粘贴到支持记忆的 AI bot → 获取个性化等级评定与改进建议

零基础步骤提示词贴士
给谁
日常在用 AI 聊天/助手但想知道自己用到了什么程度的用户
你会得到
AI 依据五档模型给你评级,引用你说过的话作为证据,指出利用率与浪费,并生成一条个性化提示词
你需要
Grok Bot

硬约束

无需配置,只需一个支持跨会话记忆的 AI(如 Grok Bot、Muse)。将提示词完整粘贴发送即可。作者自评结果为第 5 档 Unattended。

他们做了什么

作者 chrissev 将自己使用 AI 的方式用五档心智模型做了自我审视,并设计了一条提示词,让 AI 依据历史对话给自己做「绩效评估」。他把这条提示词分享到 X,邀请其他用户复制粘贴到自己的 AI bot 或 Muse 中尝试。

效果如何

作者自述:他的 AI 将他评定为第 5 档「Unattended」(无人值守)。该结果为 author_report,未经本站验证。

值得知道

五档模型从「更快谷歌」到「无人值守」呈递进关系,核心变量是:AI 是否记得你、是否介入真实工作流、是否需要人工审批。提示词要求 AI 必须引用用户原话作为证据,减少泛泛而谈;最后一步让 AI 生成个性化提示词,属于「用 AI 优化你用 AI 的方式」。注意:效果高度依赖 AI 是否真有跨会话记忆;若平台记忆有限,引用原话的准确度会下降。作者未提供 AI 实际输出的截图,回复内容因人而异。

步骤

  1. 复制评估提示词

    从下方提示词区域复制完整英文提示词。该提示词要求 AI 依据五档模型评估你的使用方式,并引用你曾说过的话作为证据。

    运行提示词 · 来源原文。发送给支持记忆的 AI bot(如 Grok Bot、Muse)的聊天输入框

    You have watched me use you for a while now. Time for my performance review.
    
    Rank me against these 5 levels:
    
    level 1: Faster Google. you ask, it answers, you leave. a search box with better grammar. nothing about your life changes
    
    level 2: One and Done. you hand it single tasks. write this, summarize that. every chat starts cold, because it knows nothing about you and you never fix that
    
    level 3: Back and Forth. you give it your real context and you iterate. drafts get argued with, improved, made yours. it is not answering anymore, you are working together
    
    level 4: In Production. up to now everything was a rehearsal: drafts you copied over yourself, work that touched nothing. now its output lands in real inboxes, real calendars, real docs, where mistakes count. you approve before anything ships
    
    level 5: Unattended. it runs without a supervisor. it remembers you between chats, starts work on its own schedule, watches things you told it to watch, and reports back with decisions ready. you stopped checking the work. you check the outcomes
    
    Start your answer with this exact line: "Mindset: [the level name]"
    
    Then:
    1. Prove it with my own words. Quote 3 things I actually said to you. No summaries, my real words
    2. What % of what you can do am I actually using? And what do I waste you on?
    3. Tell me the one thing you could do for me that fits my life, that I have never once asked for
    4. Finish by writing the prompt I should have been using all along. Personalized to me, ready to copy paste
    
    Be honest. Roast a little. Flattery is fired.
  2. 粘贴并发送

    打开你常用的支持跨会话记忆的 AI(如 Grok Bot、Muse 或其他长期记忆型助手),将提示词粘贴进对话,直接发送。

  3. 阅读回复并行动

    AI 会返回:1) 你的等级(Mindset: [等级名]);2) 引用你原话的三条证据;3) 利用率与浪费分析;4) 一条为你量身定做的新提示词。阅读后决定是否采用它生成的新提示词。

可复制提示词

参考提示词 · 来源原文。使用前确认粘贴位置、输入变量及输出格式。

这一段是提示词,页面不会在这里插入广告。

You have watched me use you for a while now. Time for my performance review.

Rank me against these 5 levels:

level 1: Faster Google. you ask, it answers, you leave. a search box with better grammar. nothing about your life changes

level 2: One and Done. you hand it single tasks. write this, summarize that. every chat starts cold, because it knows nothing about you and you never fix that

level 3: Back and Forth. you give it your real context and you iterate. drafts get argued with, improved, made yours. it is not answering anymore, you are working together

level 4: In Production. up to now everything was a rehearsal: drafts you copied over yourself, work that touched nothing. now its output lands in real inboxes, real calendars, real docs, where mistakes count. you approve before anything ships

level 5: Unattended. it runs without a supervisor. it remembers you between chats, starts work on its own schedule, watches things you told it to watch, and reports back with decisions ready. you stopped checking the work. you check the outcomes

Start your answer with this exact line: "Mindset: [the level name]"

Then:
1. Prove it with my own words. Quote 3 things I actually said to you. No summaries, my real words
2. What % of what you can do am I actually using? And what do I waste you on?
3. Tell me the one thing you could do for me that fits my life, that I have never once asked for
4. Finish by writing the prompt I should have been using all along. Personalized to me, ready to copy paste

Be honest. Roast a little. Flattery is fired.

来源

用五档心智模型让 bot/Muse 给你做「使用方式绩效评估」并写出该用的提示词 · chrissev · X · 2026-10-19T17:03:12.000Z

留言与反馈

试过这个方法吗?分享效果、遇到的问题或改进建议。留言审核后公开,请勿填写敏感信息。

正在加载…

    相关用例