Power Users/Automators 折腾/讨论/分享各种开源工具/脚本/自动化工作流👥 @zhetengsha_group📌 合集 https://t.me/zhetengsha/2🎁 恰饭 https://t.me/zhetengsha/957📢 广告投放 @xream @xream_botBuy ads: https://telega.io/c/zhetengshafeedId:55438372655431680+userId:62307599601855488http://telegram.zhetengsha.eu.orghttp://telegram.zhetengsha.eu.org/posts/5551http://telegram.zhetengsha.eu.org/posts/5551Tue, 05 May 2026 00:01:46 GMT<u>talk-normal: 让 LLM 输出更短、更直接、更像真人表达的 system prompt</u><br /><br /><s>这是第几个说人话项目了来着...</s><br /><br />• 一段通用 system prompt 即可优化 GPT、Gemini、LLaMA 等模型的回答风格，显著压缩篇幅并保留关键信息<br /><br />• 内置一套可实战迭代的表达规则，覆盖客套话清理、结论前置、条件式菜单收敛等高频输出问题<br /><br />• 提供 OpenClaw、ChatGPT 自定义指令、API 等多种接入方式，安装简单，适合个人工作流和 AI 应用集成<br /><br /><a href="https://github.com/hexiecs/talk-normal" target="_blank" rel="noopener" title="https://github.com/hexiecs/talk-normal">https://github.com/hexiecs/talk-normal</a><br /><br /><img class="tg-emoji" src="/static/https://telegram.me/i/emoji/6327880277607583303.webp" alt="" loading="lazy" width="20" height="20" /> <a href="https://t.me/zhetengsha/2" target="_blank" rel="noopener" title="合集">合集</a> <img class="tg-emoji" src="/static/https://telegram.me/i/emoji/6021860481923288796.webp" alt="" loading="lazy" width="20" height="20" /> <a href="https://t.me/zhetengsha_group" target="_blank" rel="noopener" title="群组">群组</a> <img class="tg-emoji" src="/static/https://telegram.me/i/emoji/6327792445526380187.webp" alt="" loading="lazy" width="20" height="20" /> <a href="https://t.me/zhetengsha/957" target="_blank" rel="noopener" title="恰饭">恰饭</a> <img class="tg-emoji" src="/static/https://telegram.me/i/emoji/6330367862535884142.webp" alt="" loading="lazy" width="20" height="20" /><a href="https://t.me/xream_bot" target="_blank" rel="noopener" title="联系推广">联系推广</a><br /><br /><img class="tg-emoji" src="/static/https://telegram.me/i/emoji/6327836885552991740.webp" alt="" loading="lazy" width="20" height="20" /> <a href="/search/result?q=%23LLM" title="#LLM">#LLM</a> <a href="/search/result?q=%23PromptEngineering" title="#PromptEngineering">#PromptEngineering</a> <a href="/search/result?q=%23AI" title="#AI">#AI</a> <a href="/search/result?q=%23SystemPrompt" title="#SystemPrompt">#SystemPrompt</a> <a href="/search/result?q=%23%E5%BC%80%E5%8F%91%E5%B7%A5%E5%85%B7" title="#开发工具">#开发工具</a> <a href="/search/result?q=%23%E6%95%88%E7%8E%87%E5%B7%A5%E5%85%B7" title="#效率工具">#效率工具</a> <a href="/search/result?q=%23GitHub" title="#GitHub">#GitHub</a><a class="tgme_widget_message_link_preview" href="https://github.com/hexiecs/talk-normal" target="_blank" rel="noopener" title="Make any LLM talk like a normal person. A system prompt that removes AI slop. - hexiecs/talk-normal"> <div class="link_preview_site_name accent_color">GitHub</div> <img class="link_preview_image" alt="GitHub - hexiecs/talk-normal: Make any LLM talk like a normal person. A system prompt that removes AI slop." src="/static/https://cdn4.telesco.pe/file/BaOn0rbYejjQG7HiymHXJzSU5eBBS8M9GsRRdei5mVT_PpLnpFB3W7a_mFp5X3_0z43xiAFTJc47z1XsOnE9CVmxbfqsrZ_byWluDzfLy2YCR4SVr38cliErgPkmEHzmiH3Cl3ySZhq0No-3eD7InA1dTFWX-jTZIIhev8LCgvn0yGdORa4YC13k_u01bwlcx-YeC9xfGwre7if5R_z33Gmo_KW1AuDfHwD-XM3t5vySeUQrVuIfg4ROnU1jIHT9B2ymMEAhs7pL9pG84Ml3Tg9R6K7HYFBpiJ4EijmGjd8NviHlOfM47k-PQJYvM-fruXFgI8HTBAYrG4OZH8rdHg.jpg" width="1200" height="630" loading="eager" /> <div class="link_preview_title">GitHub - hexiecs/talk-normal: Make any LLM talk like a normal person. A system prompt that removes AI slop.</div> <div class="link_preview_description">Make any LLM talk like a normal person. A system prompt that removes AI slop. - hexiecs/talk-normal</div> </a>http://telegram.zhetengsha.eu.org/posts/5354http://telegram.zhetengsha.eu.org/posts/5354Sun, 12 Apr 2026 02:00:54 GMT<u>女娲.skill: 将公开人物的思维方式蒸馏为可调用的 AI Skill</u><br /><br />• 输入一个名字即可自动完成调研、提炼、验证全流程，把心智模型、决策启发式、表达风格和价值边界沉淀为可复用的 Skill<br /><br />• 采用六路并行采集 + 三重验证机制，不是简单复读名人语录，而是提取可迁移、可预测、可解释的认知框架<br /><br />• 内置 13 位人物 Skill 和 1 个主题 Skill 示例，包含完整调研数据与效果对话，适合在 Claude Code 中直接安装、扩展和二次研究<br /><br /><a href="https://github.com/alchaincyf/nuwa-skill" target="_blank" rel="noopener" title="https://github.com/alchaincyf/nuwa-skill">https://github.com/alchaincyf/nuwa-skill</a><br /><br /><img class="tg-emoji" src="/static/https://telegram.me/i/emoji/6327880277607583303.webp" alt="" loading="lazy" width="20" height="20" /> <a href="https://t.me/zhetengsha/2" target="_blank" rel="noopener" title="合集">合集</a> <img class="tg-emoji" src="/static/https://telegram.me/i/emoji/6021860481923288796.webp" alt="" loading="lazy" width="20" height="20" /> <a href="https://t.me/zhetengsha_group" target="_blank" rel="noopener" title="群组">群组</a> <img class="tg-emoji" src="/static/https://telegram.me/i/emoji/6327792445526380187.webp" alt="" loading="lazy" width="20" height="20" /> <a href="https://t.me/zhetengsha/957" target="_blank" rel="noopener" title="恰饭">恰饭</a> <img class="tg-emoji" src="/static/https://telegram.me/i/emoji/6330367862535884142.webp" alt="" loading="lazy" width="20" height="20" /><a href="https://t.me/xream_bot" target="_blank" rel="noopener" title="联系推广">联系推广</a><br /><br /><img class="tg-emoji" src="/static/https://telegram.me/i/emoji/6327836885552991740.webp" alt="" loading="lazy" width="20" height="20" /> <a href="/search/result?q=%23AI" title="#AI">#AI</a> <a href="/search/result?q=%23%E5%BC%80%E6%BA%90%E9%A1%B9%E7%9B%AE" title="#开源项目">#开源项目</a> <a href="/search/result?q=%23%E8%AE%A4%E7%9F%A5%E8%92%B8%E9%A6%8F" title="#认知蒸馏">#认知蒸馏</a> <a href="/search/result?q=%23ClaudeCode" title="#ClaudeCode">#ClaudeCode</a> <a href="/search/result?q=%23PromptEngineering" title="#PromptEngineering">#PromptEngineering</a> <a href="/search/result?q=%23Skill" title="#Skill">#Skill</a> <a href="/search/result?q=%23%E5%BF%83%E6%99%BA%E6%A8%A1%E5%9E%8B" title="#心智模型">#心智模型</a> <a href="/search/result?q=%23%E5%86%B3%E7%AD%96%E7%B3%BB%E7%BB%9F" title="#决策系统">#决策系统</a> <a href="/search/result?q=%23GitHub" title="#GitHub">#GitHub</a><a class="tgme_widget_message_link_preview" href="https://github.com/alchaincyf/nuwa-skill" target="_blank" rel="noopener" title="你想蒸馏的下一个员工，何必是同事。蒸馏任何人的思维方式——心智模型、决策启发式、表达DNA。Distill how anyone thinks. - alchaincyf/nuwa-skill"> <div class="link_preview_site_name accent_color">GitHub</div> <img class="link_preview_image" alt="GitHub - alchaincyf/nuwa-skill: 你想蒸馏的下一个员工，何必是同事。蒸馏任何人的思维方式——心智模型、决策启发式、表达DNA。Distill how anyone thinks." src="/static/https://cdn4.telesco.pe/file/vdADVXVgRl-YRabr7TiI8a4A5LlFryKjswkeprSLYtm7C_D07btpnxjWPTbPfQbBYZ1adOXw67kQhvOEEm1UvHskmDtHD6UEbbv2glgs6KkTvmSqmYnm9aUrNztgNAfnW3Ux0zxLQ1OgsU3RNSqo__Lpq2eDQD43fW1iIp9WUXfapnGV0hY-oGPk-3omWs5X60t8YvuIGRJIz89OoNIhgmAdMN26K50oQE1UcpCQJSoX9KEXH8fwCqO_XA2VvZ2tmsH020lDr6Q0av-4KLPbMIMmBh0yaA4cEdzV7mMyiYMHMipjEX6KToUfQ3VXiKqIgADin3Wr_wOtBuvNxv1VpQ.jpg" width="1200" height="630" loading="eager" /> <div class="link_preview_title">GitHub - alchaincyf/nuwa-skill: 你想蒸馏的下一个员工，何必是同事。蒸馏任何人的思维方式——心智模型、决策启发式、表达DNA。Distill how anyone thinks.</div> <div class="link_preview_description">你想蒸馏的下一个员工，何必是同事。蒸馏任何人的思维方式——心智模型、决策启发式、表达DNA。Distill how anyone thinks. - alchaincyf/nuwa-skill</div> </a>http://telegram.zhetengsha.eu.org/posts/5090http://telegram.zhetengsha.eu.org/posts/5090Mon, 16 Mar 2026 07:02:26 GMT<div class="image-list-container image-list-odd"> <button type="button" class="image-preview-button image-preview-wrap" popovertarget="modal-5090-0" popovertargetaction="show" aria-label="Open image preview: NoPUA: 用信任驱动 AI agent 的反 PUA 调试与协作技能• 面向 Claude Code、Codex CLI、Cursor 等多种 AI agent 环境，提供可直接安装的 Skill / Rule，帮助模型在失败、卡住或反复无效尝试时自动切换到更稳健的工作模式• 保留“穷尽方案、先验证、主动排查、结构化升级”等工程方法论，同时避免用威胁式提示诱导模型隐瞒不确定性、跳过验证或过早宣称完成• 基于真实调试基准测试，重点提升隐藏问题发现能力与主动排查深度，适合用于代码调试、流水线排障、API 集成、运维与 AI 工作流优化"> <img src="/static/https://cdn5.telesco.pe/file/UHWx4kVP_2y2TLkN6v5n08H5CjpOOmDh67vIKfXJoATOoOTjS429yY1g8apj6hQaJCNWBISR9Sa8KH1BcZdO4Y9JjCZFTywcrA3op1YfHQYLEw_SYJPPt4saQLL-GvyBkoIu4DIQVn7gC6ogoZaeVrmwlMyCkIh6Mm0_iuXvpgA36VAz5CHRCbVHCVjn2LJ-OexVTt_yKGRw7F43efXeMODdonRL2S1SOmQHTzWSU59RLji3YMBkpynGW67r-Pkr9Z2LHaDg_QWk750NjmEFlGFRRnerEi74VYfWFcRBx7YzHg6AZoeqwjwgtXKWrm_BZn-ue1z9sUfST-8Du5baaw.jpg" alt="NoPUA: 用信任驱动 AI agent 的反 PUA 调试与协作技能• 面向 Claude Code、Codex CLI、Cursor 等多种 AI agent 环境，提供可直接安装的 Skill / Rule，帮助模型在失败、卡住或反复无效尝试时自动切换到更稳健的工作模式• 保留“穷尽方案、先验证、主动排查、结构化升级”等工程方法论，同时避免用威胁式提示诱导模型隐瞒不确定性、跳过验证或过早宣称完成• 基于真实调试基准测试，重点提升隐藏问题发现能力与主动排查深度，适合用于代码调试、流水线排障、API 集成、运维与 AI 工作流优化" width="800" height="433" loading="eager" /> </button> <div class="modal" id="modal-5090-0" popover="auto" aria-label="Image preview"> <button type="button" class="modal__backdrop" popovertarget="modal-5090-0" popovertargetaction="hide" aria-label="Close image preview"></button> <button type="button" class="modal__close" popovertarget="modal-5090-0" popovertargetaction="hide" aria-label="Close image preview">×</button> <div class="modal__surface"> </div> </div> </div><u>NoPUA: 用信任驱动 AI agent 的反 PUA 调试与协作技能</u><br /><br />• 面向 Claude Code、Codex CLI、Cursor 等多种 AI agent 环境，提供可直接安装的 Skill / Rule，帮助模型在失败、卡住或反复无效尝试时自动切换到更稳健的工作模式<br /><br />• 保留“穷尽方案、先验证、主动排查、结构化升级”等工程方法论，同时避免用威胁式提示诱导模型隐瞒不确定性、跳过验证或过早宣称完成<br /><br />• 基于真实调试基准测试，重点提升隐藏问题发现能力与主动排查深度，适合用于代码调试、流水线排障、API 集成、运维与 AI 工作流优化<br /><br /><a href="https://github.com/wuji-labs/nopua" target="_blank" rel="noopener" title="https://github.com/wuji-labs/nopua">https://github.com/wuji-labs/nopua</a><br /><br /><a href="/search/result?q=%23AI" title="#AI">#AI</a> <a href="/search/result?q=%23AI" title="#AI">#AI</a> agent <a href="/search/result?q=%23PromptEngineering" title="#PromptEngineering">#PromptEngineering</a> <a href="/search/result?q=%23ClaudeCode" title="#ClaudeCode">#ClaudeCode</a> <a href="/search/result?q=%23CodexCLI" title="#CodexCLI">#CodexCLI</a> <a href="/search/result?q=%23Cursor" title="#Cursor">#Cursor</a> <a href="/search/result?q=%23%E8%B0%83%E8%AF%95%E6%95%88%E7%8E%87" title="#调试效率">#调试效率</a> <a href="/search/result?q=%23%E5%B7%A5%E7%A8%8B%E5%AE%9E%E8%B7%B5" title="#工程实践">#工程实践</a> <a href="/search/result?q=%23PUA" title="#PUA">#PUA</a>