@orrisai/show-me-the-money 2.3.0 → 2.3.1

package/README.md

@@ -8,7 +8,7 @@
 [![Latest release](https://img.shields.io/github/v/release/iamzifei/show-me-the-money?label=release&color=green)](https://github.com/iamzifei/show-me-the-money/releases)
 [![License: CC BY-NC 4.0](https://img.shields.io/badge/license-CC%20BY--NC%204.0-orange.svg)](LICENSE)
 
-**Current version: `v2.3.0`** · [What's new →](#-whats-new-in-v230)
+**Current version: `v2.3.1`** · [What's new →](#-whats-new-in-v231)
 
 [English](README.md) | [中文](README.zh-CN.md)
 
@@ -51,6 +51,20 @@ Works with **Claude Code**, **Codex CLI**, **Gemini CLI**, and other agents that
 
 ---
 
+## ✨ What's New in v2.3.1
+
+**Polish patch following the v2.3.0 atom launch.** Five things landed:
+
+1. **Atom corpus grew 258 → 276** — re-ran the distill with a longer per-tweet timeout and recovered 18 of the 26 tweets that errored on the first pass (mostly long Chinese threads that needed more inference time).
+2. **Per-skill atom-loading callouts** — every non-router skill now has a Standard-startup callout at the top, declaring which atom slice it loads and reminding the agent to cite by `A-{id}` when an atom directly informs a recommendation. Previously the doctrine lived only in `skills/money/SKILL.md` and downstream skills inherited it implicitly; now it's self-documenting per file.
+3. **Workflow step ordering** — `Extract release notes` now runs **before** `Publish to npm` in the auto-publish workflow. Before this fix, a notes-extraction failure left a published-but-unreleased orphan version on npm (which is exactly what happened on the v2.3.0 ship). Notes-failure now leaves a clean rollback path.
+4. **Chinese README at full v2.3 parity** — `README.zh-CN.md` was a v1.x-era page (14 skills, no banner, no founder narrative). Rewritten to match the English version line-for-line, including the v2.1/2.2/2.3 release notes and the founder receipts.
+5. **`printf -- '...'` end-of-options fix** — already shipped in 2.3.0 but worth restating: the auto-publish workflow's `printf '----...'` separator hit bash's option-parsing on its first real run, exiting before npm publish. Now bullet-proof.
+
+No new skills, no new atoms in the corpus from sources other than #1. Library health-check only.
+
+---
+
 ## ✨ What's New in v2.3.0
 
 **The founder knowledge base ships with the suite.** Every money-* skill now auto-loads atomic principles distilled from years of solo-SaaS operating notes — battle-tested judgement embedded directly in your AI agent's working context.

package/README.zh-CN.md

@@ -1,42 +1,200 @@
+<p align="center">
+  <img src="https://pbs.twimg.com/media/HFoAbmGawAAdzJm?format=jpg&name=large" alt="Show Me The Money — 独立创业者的全自主商业操作系统" width="100%" />
+</p>
+
 # 💰 Show Me The Money
 
+[![npm version](https://img.shields.io/npm/v/@orrisai/show-me-the-money?label=npm&color=blue)](https://www.npmjs.com/package/@orrisai/show-me-the-money)
+[![Latest release](https://img.shields.io/github/v/release/iamzifei/show-me-the-money?label=release&color=green)](https://github.com/iamzifei/show-me-the-money/releases)
+[![License: CC BY-NC 4.0](https://img.shields.io/badge/license-CC%20BY--NC%204.0-orange.svg)](LICENSE)
+
+**当前版本：`v2.3.1`** · [更新内容 →](#-v231-更新内容)
+
 [English](README.md) | [中文](README.zh-CN.md)
 
-**AI 智能体技能，自主构建和运营你的商业 —— 从创意到收入，全天候运行。**
+**AI 智能体技能套件 —— 自主构建并运营你的商业，从创意到收入，全天候在线。**
+
+> 一条命令。25 个智能体技能。从零到盈利。
+
+**作者** · [X / @jamesai](https://x.com/jamesai) · [小红书 / 在悉尼和稀泥](https://www.xiaohongshu.com/user/profile/5af26425e8ac2b0a9bc030d2)
+
+---
+
+## 🎯 战绩 — 实打实的数据
+
+我是一名独立创始人，**同时在运营 6 个 SaaS 产品** —— API、开发者工具、消费级应用，全部在过去 32 个月里发布。
+
+跨这 6 个产品，当下的工作数字是：
+
+```
+💰  $1M+         全部产品的累计收入
+💳  6,500+       完成支付的交易数
+👥  10,000+      累计付费用户
+🚢  6            上线运行中的 SaaS 产品
+⏱   32 个月     从第一笔 1 美元到今天
+🧑   1 个人      没有团队、没有员工，只有 AI 智能体
+```
+
+这其中**没有一个**是用市场团队、销售部门或内容代理跑出来的。就是我一个人、一台电脑、Claude Code 终端常年开着。
+
+这个仓库里的 25 个技能，就是我用来运营自己生意的**实际操作系统**。不是课程、不是框架 —— 是我自己每天在用的可执行代码。每个技能都被用过来交付一个真实在跑、真实在 Stripe 收款的产品。
+
+**这不是空想的方法论手册。这是支撑我整个公司运转的代码。**
+
+如果你是独立创业者、indie hacker 或想以 20 人团队节奏出货的技术型创始人 —— 这是你能拿到的最接近"复制我整个工作流"的东西。
+
+---
+
+Show Me The Money 是一套开源的 [Claude Code](https://claude.ai/code) 技能套件，兼容主流 AI 编程智能体。它将你的 AI 助手变成一个全栈商业操作系统 —— 发现机会、验证需求、构建产品、运营营销、管理广告、全天候自主运行，**并且记得你在每一次会话中做出的每一个决定**。
+
+支持 **Claude Code**、**Codex CLI**、**Gemini CLI**，以及其他兼容 skill 系统的智能体。
+
+---
+
+## ✨ v2.3.1 更新内容
+
+**v2.3.0 原子发布之后的打磨补丁。** 落地 5 件事：
+
+1. **原子库 258 → 276** —— 把单条蒸馏超时从 90s 提到 180s，重跑了 26 条首次失败的推文，回收了其中 18 条（多数是长的中文 thread，需要更多推理时间）。
+2. **每个技能现在都有原子加载提示** —— 每一个非路由技能在文件顶部都有一条 Standard-startup 提示，声明它要加载哪个原子切片，并提醒 AI 在直接由原子驱动建议时用 `A-{id}` 引用。之前这条规约只写在 `skills/money/SKILL.md` 里、下游技能隐式继承；现在每个文件自包含。
+3. **发布工作流步骤顺序修复** —— 自动发布工作流里 `Extract release notes` 现在跑在 `Publish to npm` **之前**。修之前如果 notes 抽取失败，会留下一个"已发到 npm 但没有 GitHub Release"的孤儿版本（v2.3.0 上线时就遇到了这个问题）。现在 notes 失败可以干净回滚。
+4. **中文 README 完整对齐** —— `README.zh-CN.md` 之前还是 v1.x 时代的页面（14 个技能、无 banner、无战绩叙事）。已与英文版逐行对齐，包含 v2.1/2.2/2.3 的所有发布说明和创始人战绩数据。
+5. **`printf -- '...'` 修复** —— 已在 2.3.0 中发布，值得再点一下：自动发布工作流里的 `printf '----...'` 分隔符在第一次真实运行时撞上 bash 的选项解析，npm 发布前就退出了。现在防弹。
+
+没有新技能、没有从 #1 之外来源新增的原子。属于纯库健康度补丁。
+
+---
+
+## ✨ v2.3.0 更新内容
+
+**创始人原子知识库随包发布。** 每一个 money-* 技能在启动时都会自动加载从多年独立 SaaS 实战笔记中蒸馏出的原子原则 —— 把经过实战检验的判断直接嵌入 AI 助手的工作上下文。
+
+### 新增内容
+
+```
+skills/money/knowledge/atoms/
+  atoms.jsonl                              ← 完整原子库
+  atoms_solopreneur_psychology.jsonl       ← 创业者心态、行动阈值、专注度
+  atoms_market_observation.jsonl           ← 市场迁移、渠道动态、行业拐点
+  atoms_agent_infra.jsonl                  ← AI 智能体、技能设计、自动化基础设施
+  atoms_growth_tactics.jsonl               ← 外展、广告 ROI、定价、转化实验
+  atoms_content_meta.jsonl                 ← 内容策略、AI 内容陷阱、定位叙事
+```
+
+每条原子都是一句陈述性原则：一个市场观察、一个反主流判断、一个用代价换来的具体战术。蒸馏自多年的实战笔记 —— 不是观点博客。
+
+### 这意味着什么
+
+过去每次会话技能都要重新推导出同样的结论。现在 `/money-discover` 一启动就已经知道哪些市场形态对独立创始人是陷阱；`/money-content` 已经知道哪些 AI 内容模式会毁掉品牌定位；`/money-diagnose` 在用户的失败模式匹配到已知原子时会直接引用它 —— 任何建议都可以追溯回原始观察。
+
+当某条建议直接由原子驱动时，技能会用 ID 引用它：
+
+> "在这里选 $29/月的消费级切入点会踩中 **A-bce2** 的陷阱 —— Agent 基础设施正在 12 个月内把消费 App 推向无 UI 的纯 API 形态。"
+
+点开来源链接你能看到原始观察、日期、原始上下文。
+
+### 现在有了两层记忆
+
+- **原子（本次发布）** —— 全局、由创始人维护、随包发布、运行时只读。承载通用原则。
+- **学习（v2.2 引入）** —— 项目级、按 slug 自动捕获、可变。承载本项目特有的模式。
+
+合在一起：每个技能启动时既有**通用经营智慧**，也有**对这个特定生意我们已经学到的东西**。
+
+### 维护者侧：增量蒸馏
+
+产出原子的管线（`scripts/x-distill.mjs`，gitignored，仅维护者可见）是增量的 —— 后续运行只处理新内容，所以语料库会自然生长，无需重跑全档。
+
+---
+
+## ✨ v2.2.0 更新内容
+
+**评审委员会 + 跨会话学习。** 8 个新技能，加上把过往洞见自动加载进每一次未来会话的能力。
+
+### 计划评审擂台
+
+大多数独立创始人会把自己的计划自我说服成"绿灯通行"，然后跑 6 个月才发现结构性缺陷。现在有 4 个独立评审角色 —— 各自完整人格 —— 加 1 个总编排：
+
+```
+/money-review-investor   ← 投资人模式：SEED VIABLE / LATER ROUND / BOOTSTRAP-ONLY / UNFUNDABLE
+/money-review-customer   ← 指定 ICP 模式：PAY NOW / WITH FRICTION / WRONG POSITIONING / WRONG ICP
+/money-review-operator   ← 单人执行模式：SHIPPABLE / DESCOPE / NEEDS HIRE / WRONG STACK
+/money-review-skeptic    ← 质疑者：EXISTENTIAL / SOLVABLE / LOW-RISK / WRONG QUESTION
+/money-panel             ← 一次跑齐 4 个，找一致项，只把品味分歧抛给你
+```
+
+一条命令跑完整套擂台，只在边界判断处停下来等你拍板。
+
+### 跨会话学习，终于落地
+
+```
+/money-learn       ← 原子化、被验证过的模式，自动注入每个其他技能
+/money-retro       ← 周复盘：决定 / 出货 / 卡住 / 未用技能 / 焦点
+/money-skillify    ← 把成功的工作流固化为项目级技能
+```
+
+每个 money-* 技能现在都会在启动时自动加载相关学习。AI 在一次次会话中越来越懂你的项目，而不是每次冷启动从零开始。
+
+### /money-diagnose 的铁律（Iron Law）
 
-> 一条命令。14 个智能体技能。从零到盈利。
+`/money-diagnose` 现在按 **4 个明确阶段**运行（调查 → 分析 → 假设 → 建议），并在第 3 阶段设硬门：**未经用户明确确认根因假设之前，不允许给建议。** 这堵住了诊断技能最常见的失败模式 —— AI 80% 自信、用户客气地"嗯对"了一下、然后 30 分钟的建议全部射偏目标。
 
-Show Me The Money 是一套开源的 [Claude Code](https://claude.ai/code) 技能套件，兼容主流 AI 编程智能体。它将你的 AI 助手变成一个全栈商业操作系统 —— 发现机会、验证需求、构建产品、运营营销、管理广告，并全自主运行。
+技能里有可选的 Claude Code 钩子，做工具级强制，文档已附。
 
-支持 **Claude Code**、**Codex CLI**、**Gemini CLI** 及其他兼容智能体。
+---
 
-## 快速开始
+## ✨ v2.1.0 更新内容
+
+**跨会话状态管理。** 三个技能 —— `/money-save`、`/money-restore`、`/money-report` —— 把每一次对话变成一个检查点。上周敲定的定价、被排除的方向、正在测的假设，全部跨 Claude Code 会话留存。
+
+```
+session 1  →  /money-discover  →  /money-save     ✅ 切入点已锁定
+   …一周过去…
+session 2  →  /money-restore   →  从这里继续        📦 不需要重新解释
+           →  /money-strategy  →  /money-save     ✅ 定价已锁定
+   …一个月过去…
+session N  →  /money-report                        📄 全量交付物
+```
+
+---
+
+## 🚀 快速开始
+
+### 方式 1 — Claude Code 插件市场
+
+```bash
+claude plugin marketplace add iamzifei/show-me-the-money
+claude plugin install money@show-me-the-money
+```
+
+### 方式 2 — npx（任意智能体均可）
 
 ```bash
 npx @orrisai/show-me-the-money
 ```
 
-然后打开 Claude Code 并输入：
+然后打开 Claude Code（或 Codex CLI / Gemini CLI）输入：
 
 ```
 /money
 ```
 
-就这样。AI 会自动引导你，构建你的画像，并带你完成每一步。
+就这样。AI 会自动检查是否有先前会话状态、为新用户做引导、带你完成每一步。
+
+---
 
-## 功能概览
+## 🧭 它能做什么
 
 ```
-📧 入职引导          从公开数据构建你的用户画像
+📧 入职引导          基于公开数据构建你的画像
        │
        ▼
-💡 发现商机          扫描市场，验证需求，找到切入点
+💡 发现商机          扫描市场、验证需求、找到你的切入点
        │
        ▼
-📊 制定策略          市场研究，SWOT，4P，商业模式，GTM 方案
+📊 制定策略          市场研究、SWOT、4P、商业模式、GTM 方案
        │
        ▼
-🔨 构建产品          发布 MVP：落地页、支付、SEO 优化
+🔨 构建产品          落地页、支付、SEO/GEO 一键 MVP
        │
        ▼
 📈 增长引擎          内容、社交、外展、SEO、广告 —— 全自动
@@ -45,46 +203,78 @@ npx @orrisai/show-me-the-money
 🤖 全天候运营         自主运营、监控、财务报告
        │
        ▼
+💾 持久化            /money-save → /money-restore → /money-report
+       │
+       ▼
 💵 收入
 ```
 
 ### 完整流水线
 
 1. **入职引导** — 分享邮箱和社媒账号，AI 自动调研你的背景，构建个性化画像
-2. **发现** (`/money-discover`) — 扫描市场空白，用 6 个验证问题验证需求，5 层过滤评估 + 竞品情报协议
-3. **策略** (`/money-strategy`) — 前提解构协议 + 完整市场研究报告：SWOT、4P、商业模式压力测试、GTM 方案
-4. **诊断** (`/money-diagnose`) — 深度商业诊断：问题解构漏斗、假设审计、执行力辅导
-5. **产品** (`/money-product`) — 构建并部署 MVP：落地页、认证、支付、SEO/GEO、QA 测试、金丝雀监控
-6. **质量** (`/money-quality`) — 上线前质量门：代码审查、安全审计、性能检查、可访问性
-7. **内容** (`/money-content`) — 博客、邮件、社交、视频脚本 — 经过 5 维诊断 + 真实性审计 + 标题影响力矩阵
-8. **外展** (`/money-outreach`) — 冷邮件序列、合作伙伴拓展、个性化潜客开发
-9. **社交** (`/money-social`) — 跨 X、LinkedIn、Reddit、Product Hunt 社交媒体管理，配合 Hook 写作框架
-10. **SEO** (`/money-seo`) — 传统 SEO + GEO（ChatGPT、Perplexity、Gemini 的 AI 搜索优化）
-11. **广告** (`/money-ads`) — Google Ads、Meta Ads — 投放设置、优化、ROAS 追踪
-12. **运营** (`/money-ops`) — 全天候自主运营 + 商业健康评分 + 安全护栏
-13. **财务** (`/money-finance`) — 收入追踪、单位经济、财务报告
-
-## 为什么做这个
-
-大多数 AI 商业工具只给你泛泛的建议。Show Me The Money 给你**可执行的工作流** —— 每个技能都产出具体行动，而不仅仅是分析。每个阶段都以"明天的第一个行动：[具体任务]"结尾。
+2. **发现** (`/money-discover`) — 扫描市场空白，6 个验证问题确认需求，5 层过滤评估 + 竞品情报协议
+3. **策略** (`/money-strategy`) — 前提解构 + 完整市场研究报告：SWOT、4P、商业模式压力测试、GTM 方案
+4. **产品** (`/money-product`) — 构建并部署 MVP：落地页、认证、支付、SEO/GEO、QA、金丝雀监控
+5. **质量** (`/money-quality`) — 上线前质量门：代码审查、QA 测试、安全审计、性能、可访问性
+6. **内容** (`/money-content`) — 博客、邮件序列、社交、视频脚本，配 5 维质量诊断 + 12 信号真实性审计 + 标题影响力矩阵
+7. **外展** (`/money-outreach`) — 冷邮件序列、合作拓展、个性化潜客
+8. **社交** (`/money-social`) — X / LinkedIn / Reddit / Product Hunt 跨平台运营，Hook 写作框架加持
+9. **SEO** (`/money-seo`) — 传统 SEO + GEO（ChatGPT、Perplexity、Gemini 等 AI 搜索优化）
+10. **广告** (`/money-ads`) — Google Ads、Meta Ads —— 投放设置、优化、ROAS 追踪
+11. **运营** (`/money-ops`) — 全天候自主运营 + 商业健康评分 + 金丝雀监控 + 安全护栏
+12. **财务** (`/money-finance`) — 收入追踪、单位经济、财务报告
+13. **诊断** (`/money-diagnose`) — 业务卡住时深度诊断：问题解构、假设审计、执行力辅导，带 Iron Law 阶段门
+14. **保存 / 恢复 / 报告** (`/money-save`、`/money-restore`、`/money-report`) — 跨会话状态管理：检查点 + 续接 + 多月进展合并成一份可分享的交付物
+
+---
+
+## 🌟 为什么做这个
+
+大多数 AI 商业工具只给你泛泛的建议。Show Me The Money 给你**可执行的工作流** —— 每个技能都产出具体行动，不只是分析。每个阶段都以"明天的第一个行动：[具体任务]"收尾。
 
 **核心差异：**
-- **个性化** — 自动调研你的背景，所有建议都针对你量身定制
-- **验证框架** — 集成实战方法论：前提解构协议、6 问需求验证、5 层机会评分、商业模式压力测试、5 维内容诊断、12 信号真实性审计、标题影响力矩阵
-- **全栈覆盖** — 覆盖整个商业生命周期，而不仅仅是某一环
-- **收入导向** — 每条建议都与赚钱挂钩，没有废话
-- **自主运行** — 设置一次，通过 `/money-ops` 全天候运行
-- **开源** — CC BY-NC 4.0 许可证。个人使用免费。自定义、扩展、贡献
 
-## 安装
+- **个性化** — 自动调研你的背景，所有建议针对你量身定制
+- **被验证过的框架** — 实战方法论：前提解构、6 问需求验证、5 层机会评分、商业模式压力测试、5 维内容诊断、12 信号真实性审计、标题影响力矩阵
+- **全栈覆盖** — 覆盖完整商业生命周期，不止某一环
+- **收入导向** — 每条建议都连回赚钱，没有废话
+- **持久化记忆** — 决策、被排除的方向、正在测的假设，通过 `/money-save` 跨会话留存
+- **创始人级判断** — 通过 `/money-learn`（项目级）+ 创始人原子库（全局），AI 越用越像你的脑子
+- **自主运行** — 一次配置，通过 `/money-ops` 全天候运行
+- **开源** — CC BY-NC 4.0 协议。个人使用免费。可定制、可扩展、欢迎贡献
+
+---
 
-### 通过 npx（推荐）
+## 📦 安装
+
+### 通过 Claude Code 插件市场（Claude Code 用户推荐）
+
+```bash
+claude plugin marketplace add iamzifei/show-me-the-money
+claude plugin install money@show-me-the-money
+```
+
+之后更新：
+
+```bash
+claude plugin marketplace update show-me-the-money
+claude plugin update money@show-me-the-money
+/reload-plugins
+```
+
+### 通过 npx（任意智能体均可 —— Claude Code、Codex、Gemini）
 
 ```bash
 npx @orrisai/show-me-the-money
 ```
 
-自动将全部 14 个技能安装到 `~/.claude/skills/`。
+自动将全部 25 个技能安装到 `~/.claude/skills/`。
+
+之后更新，跑同一条命令即可：
+
+```bash
+npx @orrisai/show-me-the-money
+```
 
 ### 通过 npm（全局安装）
 
@@ -99,34 +289,63 @@ git clone https://github.com/iamzifei/show-me-the-money.git ~/.claude/skills/sho
 cd ~/.claude/skills/show-me-the-money && node install.js
 ```
 
-### 更新
+---
 
-```bash
-npx @orrisai/show-me-the-money update
-```
+## 🛠 技能列表
 
-自动检查最新版本、下载并重新安装所有技能。
+### 核心（路由、状态、学习）
+
+| 技能 | 命令 | 功能 |
+|------|------|------|
+| **路由器** | `/money` | 用户引导、画像构建、检查先前状态、智能信号路由 |
+| **保存** | `/money-save` | 把当前业务状态检查点写入 `~/.smtm/sessions/{project}/` |
+| **恢复** | `/money-restore` | 从先前的检查点续接，无需重新解释 |
+| **报告** | `/money-report` | 把所有保存的状态合并成可交付的 markdown 报告 |
+| **学习** | `/money-learn` | 管理项目级原子学习（自动注入每个其他技能） |
+| **固化技能** | `/money-skillify` | 把一段成功的工作流固化为项目级可复用技能 |
+| **升级** | `/money-upgrade` | 更新到最新版本 |
 
-## 技能列表
+### 发现 · 策略 · 诊断 · 评审
 
-| 技能 | 命令 | 功能说明 |
-|------|------|----------|
-| **路由器** | `/money` | 用户引导、画像构建、智能信号路由 |
+| 技能 | 命令 | 功能 |
+|------|------|------|
 | **发现** | `/money-discover` | 发现并验证盈利商机 + 竞品情报 |
 | **策略** | `/money-strategy` | 前提解构 + 市场研究 + 商业模式压力测试 |
-| **诊断** | `/money-diagnose` | 深度商业诊断：根因分析 + 执行力辅导 |
-| **产品** | `/money-product` | 构建、部署、QA 测试、金丝雀监控 |
+| **诊断** | `/money-diagnose` | 业务卡住时找根因，不是症状（带 Iron Law 阶段门） |
+| **评审委员会** | `/money-panel` | 4 个评审一起跑，只把品味分歧抛给你 |
+| **投资人评审** | `/money-review-investor` | VC 视角的融资可行性评判 |
+| **客户评审** | `/money-review-customer` | 指定 ICP 视角的付费意愿评判 |
+| **操盘手评审** | `/money-review-operator` | 独立创始人执行可行性评判 |
+| **质疑者评审** | `/money-review-skeptic` | 红队式质疑评审 |
+
+### 构建 · 质量
+
+| 技能 | 命令 | 功能 |
+|------|------|------|
+| **产品** | `/money-product` | 构建、部署、QA、监控 MVP |
 | **质量** | `/money-quality` | 代码审查、安全审计、性能检查、上线前门禁 |
+
+### 增长
+
+| 技能 | 命令 | 功能 |
+|------|------|------|
 | **内容** | `/money-content` | 内容管线 + 真实性审计 + 标题影响力矩阵 |
 | **外展** | `/money-outreach` | 冷邮件序列、合作伙伴拓展 |
 | **社交** | `/money-social` | 社交媒体管理 + Hook 写作框架 |
-| **SEO** | `/money-seo` | SEO + GEO 优化 |
-| **广告** | `/money-ads` | Google Ads、Meta Ads — 设置、优化、ROAS |
+| **SEO** | `/money-seo` | SEO + GEO 优化（含 AI 搜索） |
+| **广告** | `/money-ads` | Google Ads、Meta Ads —— 设置、优化、ROAS |
+
+### 运营 · 复盘
+
+| 技能 | 命令 | 功能 |
+|------|------|------|
 | **运营** | `/money-ops` | 全天候运营 + 健康评分 + 安全护栏 |
-| **财务** | `/money-finance` | 收入追踪和财务报告 |
-| **升级** | `/money-upgrade` | 更新到最新版本 |
+| **财务** | `/money-finance` | 收入追踪、财务报告 |
+| **复盘** | `/money-retro` | 周复盘：决定 / 出货 / 卡住 / 未用技能 / 焦点 |
 
-## 使用示例
+---
+
+## 💡 使用示例
 
 ### 从零开始
 ```
@@ -161,58 +380,101 @@ npx @orrisai/show-me-the-money update
 /money-ops "为我在 example.com 的产品自动化内容、社交和 SEO"
 ```
 
-## 兼容性
+### 锁定决策，下次继续
+
+```
+/money-discover "..."
+→ AI 验证切入点，你确认定价
+/money-save                       ✅ 切入点 + 定价已检查点
+
+   …一周后，新的 Claude Code 会话…
+
+/money-restore                    📦 从你上次离开的地方继续
+/money-strategy                   → 在此基础上构建 GTM
+```
+
+### 给联合创始人生成交付物
+
+```
+/money-report
+→ 把你跨数周或数月保存的所有检查点合并成一份
+   可分享的 markdown 报告，输出到 ~/.smtm/reports/{project}/
+```
+
+---
+
+## 🌐 兼容性
 
-Show Me The Money 兼容任何支持 `~/.claude/skills/` 技能系统的 AI 编程智能体：
+Show Me The Money 兼容任何支持 `~/.claude/skills/` 技能系统或 Claude Code 插件市场的 AI 编程智能体：
 
-- **Claude Code**（首选）
+- **Claude Code**（首选，完整插件市场支持）
 - **Codex CLI**
 - **Gemini CLI**
 - **Cursor**（通过 skills）
 - **其他兼容智能体**
 
-## 工作原理
+---
 
-基于 [Claude Code 技能系统](https://docs.anthropic.com/en/docs/claude-code/skills) 构建。每个技能是一个 `SKILL.md` 文件，包含分步工作流、商业框架和决策树，AI 自主遵循执行。
+## 🏗 工作原理
+
+基于 [Claude Code 技能系统](https://docs.anthropic.com/en/docs/claude-code/skills) 构建。每个技能是一个 `SKILL.md` 文件，包含分步工作流、商业框架、决策树，AI 自主遵循执行。
 
 ```
 ~/.claude/skills/
-├── money/SKILL.md              ← 路由器 + 用户引导 + 智能信号路由
-├── money-discover/SKILL.md     ← 创意发现 + 竞品情报
-├── money-strategy/SKILL.md     ← 前提解构 + 市场研究
-├── money-diagnose/SKILL.md     ← 商业诊断 + 执行力辅导
-├── money-product/SKILL.md      ← 产品构建 + QA + 金丝雀监控
-├── money-quality/SKILL.md      ← 代码审查 + 安全 + 性能门禁
-├── money-content/SKILL.md      ← 内容管线 + 真实性审计
-├── money-outreach/SKILL.md     ← 销售与外展
-├── money-social/SKILL.md       ← 社交媒体
-├── money-seo/SKILL.md          ← SEO 与 GEO
-├── money-ads/SKILL.md          ← 付费广告
-├── money-ops/SKILL.md          ← 全天候运营 + 健康评分
-├── money-finance/SKILL.md      ← 财务追踪
-└── money-upgrade/SKILL.md      ← 版本管理
-```
-
-## 卸载
+├── money/SKILL.md                ← 路由器 + 入职引导 + 先前状态检查
+├── money/knowledge/atoms/        ← 创始人原子知识库（v2.3 起随包发布）
+├── money-discover/SKILL.md       ← 创意发现 + 竞品情报
+├── money-strategy/SKILL.md       ← 前提解构 + 市场研究
+├── money-diagnose/SKILL.md       ← 业务诊断 + 执行力辅导（含 Iron Law）
+├── money-panel/SKILL.md          ← 4 评审编排器
+├── money-review-{investor,customer,operator,skeptic}/SKILL.md
+│                                 ← 4 个独立评审角色
+├── money-product/SKILL.md        ← 产品构建 + QA + 金丝雀
+├── money-quality/SKILL.md        ← 代码审查 + 安全 + 性能门禁
+├── money-content/SKILL.md        ← 内容管线 + 真实性审计
+├── money-outreach/SKILL.md       ← 销售与外展
+├── money-social/SKILL.md         ← 社交媒体
+├── money-seo/SKILL.md            ← SEO 与 GEO
+├── money-ads/SKILL.md            ← 付费广告
+├── money-ops/SKILL.md            ← 全天候运营 + 健康评分
+├── money-finance/SKILL.md        ← 财务追踪
+├── money-save/SKILL.md           ← 跨会话检查点写入
+├── money-restore/SKILL.md        ← 跨会话状态加载
+├── money-report/SKILL.md         ← 交付物报告生成
+├── money-learn/SKILL.md          ← 项目级学习管理
+├── money-retro/SKILL.md          ← 周复盘
+├── money-skillify/SKILL.md       ← 工作流固化为项目级技能
+└── money-upgrade/SKILL.md        ← 版本管理
+```
+
+状态文件落在 `~/.smtm/sessions/{project}/`（按项目、追加写）和 `~/.smtm/reports/{project}/`（带时间戳、永不覆写）。学习落在 `~/.smtm/projects/{project}/learnings.jsonl`。原子库随包发布，运行时只读。
+
+---
+
+## 🗑 卸载
 
 ```bash
 npx @orrisai/show-me-the-money uninstall
 ```
 
-## 贡献
+---
+
+## 🤝 贡献
 
 1. Fork 本仓库
 2. 创建功能分支
-3. 编辑 `skills/` 中的 SKILL.md 文件
+3. 编辑 `skills/` 中的 `SKILL.md` 文件
 4. 复制到 `~/.claude/skills/` 并在 Claude Code 中测试
 5. 提交 PR
 
-欢迎所有贡献 — 新技能、框架改进、语言支持、Bug 修复。
+欢迎所有贡献 —— 新技能、框架改进、语言支持、Bug 修复。
+
+---
 
-## 许可证
+## 📄 许可证
 
 CC BY-NC 4.0（知识共享 署名-非商业性使用 4.0）
 
-- 个人/学习/研究：自由使用
+- 个人 / 学习 / 研究：自由使用
 - 公开衍生作品：需注明来源
 - 商业用途：需单独授权

package/VERSION

@@ -1 +1 @@
-2.3.0
+2.3.1

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@orrisai/show-me-the-money",
-  "version": "2.3.0",
+  "version": "2.3.1",
   "description": "AI agent skills that build and run your business autonomously — from idea to revenue, 24/7. Works with Claude Code, Codex CLI, and Gemini CLI.",
   "keywords": [
     "ai-agent",

package/skills/money/knowledge/atoms/atoms.jsonl

@@ -256,3 +256,21 @@
 {"id":"A-aac4","captured_at":"2026-05-03T08:17:07.906Z","source":"https://x.com/jamesai/status/1750690968771461332","source_text":"独立开发，重点是在独立而不是在开发。\n\n看到了多年前的自己。感觉这个问题挺典型的。\n\n#indie #indiehackers #独立开发 https://t.co/ibfCQYpMAr","source_created_at":"2024-01-26T01:23:47.000Z","category":"solopreneur_psychology","pattern":"独立开发的核心在于「独立」（自主决策、不依附体制），而非「开发」（技术执行）本身。","confidence":"emerging","tags":["独立开发","心态重塑","indie"],"distill_model":"sonnet","reason":"该推文包含一个可泛化的宣言式原则：对「独立开发」概念的重新定义——将核心从技术执行转移到自主性本身，具有典型的创始人心理洞察价值。"}
 {"id":"A-d0f6","captured_at":"2026-05-03T08:18:01.969Z","source":"https://x.com/jamesai/status/1732655724780806614","source_text":"就这样莫名其妙的mrr破了4000。\n\n这2个多月，尤其是最近几个礼拜，随着社区和社媒越来越多，得到的信息和方法论也越来越多。按道理应该是好事，但是结果是更慢的交付和更焦虑的心态。工程师，学院派和方法论，这些叠加到一起，真能提高成功的概率吗？\n\n年底上班事也多，心累… https://t.co/bTnLgaKhHQ","source_created_at":"2023-12-07T06:58:10.000Z","category":"solopreneur_psychology","pattern":"信息和方法论积累越多，反而导致交付更慢、心态更焦虑——工程师叠加学院派思维，未必能提高创业成功概率。","confidence":"emerging","tags":["方法论陷阱","信息过载","执行速度"],"distill_model":"sonnet","reason":"包含明确的反主流洞察：社区与方法论的积累按理是好事，但创始人亲历结果反而是更慢交付与更高焦虑，构成可推广的 solopreneur 认知陷阱。"}
 {"id":"A-00b3","captured_at":"2026-05-03T08:19:37.486Z","source":"https://x.com/jamesai/status/1725376248267059565","source_text":"最近才开始兼职做点独立开发，提醒自己切记别陷入独立开发的茧房。\n\n从自身和独立开发的小需求出发，在加上独立开发社群的互助，确实能够解决0-1而且能快速有正反馈。但是从破圈，上限和变现周期的角度，可能还不如上班或者做内容简单粗暴😂","source_created_at":"2023-11-17T04:52:08.000Z","category":"solopreneur_psychology","pattern":"独立开发社群能帮助快速完成 0→1 并获得正反馈，但其\"茧房\"效应会压低变现天花板，破圈速度和变现周期可能不如做内容或上班。","confidence":"emerging","tags":["独立开发茧房","变现天花板","社群回音壁"],"distill_model":"sonnet","reason":"该推文包含一个可泛化的反主流洞察：独立开发社群提供的正反馈循环是一种认知茧房，对破圈和变现速度有结构性限制，适合提炼为创始人心理模型原子。"}
+{"id":"A-17dc","captured_at":"2026-05-03T09:55:32.580Z","source":"https://x.com/jamesai/status/2034926502836674659","source_text":"玩了 2 天的 OpenClaw。我感觉很多人弄了一堆 Agent 甚至是什么“三省六部”，感觉更多的是叙事？我用下来没有应用场景啊。\n\nAI Agent 从来不是输入和输出的瓶颈，我才是瓶颈。\n\n对我来说甚至没办法做到 OpenClaw 在我离线的时候持续工作，更别说还要弄一堆 Agent 持续分工合作了。","source_created_at":"2026-03-20T09:34:20.000Z","category":"agent_infra","pattern":"AI Agent 从来不是输入输出的瓶颈，创始人自己才是——多 Agent 架构的繁荣更多是叙事价值而非实用价值。","confidence":"emerging","tags":["multi-agent","human-bottleneck","agent-narrative"],"distill_model":"sonnet","reason":"包含一个反主流宣言并附有推理：复杂多 Agent 架构的限制不在模型能力，而在于人类创始人无法持续提供上下文与方向，由此推出\"三省六部\"式 Agent 编排更多是叙事噱头而非实际生产力杠杆。"}
+{"id":"A-3952","captured_at":"2026-05-03T09:56:04.558Z","source":"https://x.com/jamesai/status/2008729167102431270","source_text":"刷推刷多了，天天都在获取新的信息和知识。\n\n在出现自己认知提高的“幻觉”同时，然后就觉得自己离钱其实越来越远了。\n\n还是要多接触土老板和个体户。","source_created_at":"2026-01-07T02:35:28.000Z","category":"solopreneur_psychology","pattern":"刷信息流制造\"认知提升\"幻觉，同时让你离真实的钱越来越远；赚钱的洞察藏在土老板和个体户身上，而非内容算法里。","confidence":"hypothesis","tags":["information trap","grassroots wisdom","attention allocation"],"distill_model":"sonnet","reason":"包含明确的反主流观点与因果推断：信息消费→认知幻觉→远离金钱，并给出可操作替代方案（接触土老板/个体户），构成完整的可复用原则。"}
+{"id":"A-84cf","captured_at":"2026-05-03T09:57:01.687Z","source":"https://x.com/jamesai/status/1908684183129776256","source_text":"《谷歌广告投放入门》\n\n虽然很基础，但是觉得对新手来说也足够了。\n\n我之前就是靠这个拿到的第一桶金，剩下的就是经验和迭代了。\n\n所以说赚钱并不一定要是专家，大胆去实践才是王道 https://t.co/vK2knHU9nW","source_created_at":"2025-04-06T00:52:25.000Z","category":"solopreneur_psychology","pattern":"赚到第一桶金不需要专家级认知，降低行动阈值、大胆实践比系统学习更关键。","confidence":"validated","tags":["行动阈值","入门即实践","第一桶金"],"distill_model":"sonnet","reason":"推文包含一个可泛化的创业心理原则：用亲身赚钱经历验证了\"够用的知识＋立即行动\"优于\"等成为专家再出手\"，具备可复用的规律性主张。"}
+{"id":"A-05da","captured_at":"2026-05-03T09:57:46.035Z","source":"https://x.com/jamesai/status/1905921193502486923","source_text":"很多人都在卷 seo，但以后的用户，可能根本不是从那儿来的。\n\n现在一部分人搜索工具，第一步不是搜谷歌百度，也不是查知乎，而是直接问 ChatGPT、DeepSeek、Perplexity。\n\n问题来了：怎么让 AI 把你当成“推荐项”？👇","source_created_at":"2025-03-29T09:53:17.000Z","category":"market_observation","pattern":"用户搜索工具的第一入口正在从谷歌/百度迁移到 ChatGPT/Perplexity，SEO 的重要性让位给\"如何让 AI 把你列为推荐项\"（GEO）。","confidence":"emerging","tags":["GEO","AI搜索","渠道迁移"],"distill_model":"sonnet","reason":"该推文包含一个明确的渠道迁移观察——用户搜索起点正在从传统搜索引擎转向 AI 对话工具——并以此引出可操作的战略问题，具有清晰的市场含义，可提炼为独立原子。"}
+{"id":"A-d987","captured_at":"2026-05-03T09:58:24.780Z","source":"https://x.com/jamesai/status/1904016723717874039","source_text":"AI不是镰刀，它只是更换了手柄的锤子。\n\n以前靠卖力气干活，有了 AI 如果只是想着省力气干活。那一辈子就在干活 😂。\n\n现在拼的是有没有一张“关系网”：\n🧱没地没矿=没有生产资料；\n😶‍🌫️没粉没圈=没有生产关系；\n💻只靠写代码、搬砖=纯纯的生产力，用不用 AI 都一样。","source_created_at":"2025-03-24T03:45:36.000Z","category":"solopreneur_psychology","pattern":"AI 只改变工具效率，不改变竞争结构：没有粉丝/社群（生产关系）和资产/渠道（生产资料）的纯技术人，用不用 AI 结局一样。","confidence":"emerging","tags":["AI杠杆","流量护城河","竞争结构"],"distill_model":"sonnet","reason":"Tweet makes a declarative, generalizable claim using a Marxist production framework to argue that AI doesn't redistribute power — distribution (followers/community) and assets remain the decisive moats, not raw technical productivity."}
+{"id":"A-b568","captured_at":"2026-05-03T09:59:36.482Z","source":"https://x.com/jamesai/status/1903330769927586055","source_text":"很多时候人赚不到钱，就是因为还不够穷。\n\n我曾经对外包嗤之以鼻，但是多年以后，我还是那个背着房贷的牛马，而身边几个做外包的小伙伴，有的已经转型做咨询和培训，有的不光有稳定的业务，而且已经攒了一大堆开箱即用的“轮子”，收客户定制的钱，干的是换皮的活，生意越做越轻松了。 https://t.co/DOKi4eoxaF","source_created_at":"2025-03-22T06:19:52.000Z","category":"solopreneur_psychology","pattern":"对外包嗤之以鼻是一种昂贵的身份偏见——做外包能积累可复用\"轮子\"、降低边际成本、转型咨询，比死守高地位路径更快实现财务自由。","confidence":"emerging","tags":["status-bias","outsourcing","asset-accumulation"],"distill_model":"sonnet","reason":"包含反主流观点+亲身对比案例：鄙视外包的人原地踏步，做外包的人积累轮子后越做越轻松，揭示了\"身份偏见阻碍商业行动\"这一可泛化的创业心理规律。"}
+{"id":"A-efe4","captured_at":"2026-05-03T10:00:20.245Z","source":"https://x.com/jamesai/status/1902513306423980537","source_text":"在今年 6 月 17 号之后，Stripe 每收到一笔 dispute（争议性付款）后，商家除了要支付 15 美金的费用外，如果有异议，还要额外再付 15 美金，最后如果赢了会退还这 15 美金。\n\n也就是说，如果现在有 fraud 用黑卡刷了你的服务，或者哪怕是一些 SB 用户付完钱后直接 dispute。你就要付 30 美金，且里面 https://t.co/jJ67dewsWF","source_created_at":"2025-03-20T00:11:33.000Z","category":"market_observation","pattern":"Stripe 新 dispute 费用政策（争议最高收 30 美金）大幅提升欺诈成本，SaaS 独立开发者若不主动做反欺诈拦截，黑卡刷单和恶意 chargeback 将直接侵蚀利润。","confidence":"validated","tags":["Stripe","chargeback","fraud-cost"],"distill_model":"sonnet","reason":"明确的平台政策变化+对 SaaS 创业者的直接财务影响，属于有具体数字支撑的市场观察，值得独立提炼为风控注意事项。"}
+{"id":"A-422d","captured_at":"2026-05-03T10:02:03.381Z","source":"https://x.com/jamesai/status/1901413433075274219","source_text":"雷军的营销或者所谓创始人 IP 太有迷惑性了，我时不时就要重刷一遍雷军十几年前的演讲和采访，提醒一下自己成功的路径，本质上是长期积累、深度思考的结果，而不是单纯的营销技巧。\n\n很多人看雷军学营销，就像直接抄袭一个成熟产品的 UI 设计，而不去理解整个产品是如何一步步走到这一步的。","source_created_at":"2025-03-16T23:21:03.000Z","category":"solopreneur_psychology","pattern":"学习成功创始人应追溯其长期积累与深度思考的底层路径，而非模仿其表层营销技巧——直接复制结果而忽视过程，如同抄 UI 而不懂产品演化。","confidence":"emerging","tags":["创始人IP","第一性原理","学习方法"],"distill_model":"sonnet","reason":"包含一个可泛化的创始人认知原则：区分\"结果层模仿\"与\"路径层学习\"，并用 UI 抄袭类比强化论点，属于有明确推理的 solopreneur 心智模型。"}
+{"id":"A-a97e","captured_at":"2026-05-03T10:02:39.044Z","source":"https://x.com/jamesai/status/1900899544784875822","source_text":"kolfind 这 1-2 个月我基本上没做什么主动的运营动作。推上我聊的很少，就默默的做 SEO。作为初学者，速度很慢，但是流量慢慢在上涨，这个月已经破万了，比上个月上涨了 240+%。\n\n我从上线后就基本上没再刻意去“带货”它，这很不 buildinpublic。我不能靠follower的支持来让我对做产品产生误判。 https://t.co/d8Kt6ZCtQN","source_created_at":"2025-03-15T13:19:03.000Z","category":"solopreneur_psychology","pattern":"Build-in-public 的 follower 支持会制造虚假的产品验证信号，有机 SEO 增长才是判断产品真实可行性的可靠指标。","confidence":"emerging","tags":["build-in-public","product-validation","SEO"],"distill_model":"sonnet","reason":"包含一个反主流原则（刻意回避 build-in-public）及其背后的认知逻辑（follower 支持 ≠ 真实产品需求），并以具体数据（流量破万、240% 增长）支撑，可泛化为独立创始人的产品验证方法论。"}
+{"id":"A-85f1","captured_at":"2026-05-03T10:03:09.688Z","source":"https://x.com/jamesai/status/1899052617436729449","source_text":"副业听起来“既有又有”很美好，但是人真的不能一直搞副业。不是说副业不好，而是一直搞副业，可能会让自己永远停留在“边缘地带”。","source_created_at":"2025-03-10T11:00:01.000Z","category":"solopreneur_psychology","pattern":"长期停留在副业模式会让创业者永远处于\"边缘地带\"，无法真正进入主场。","confidence":"hypothesis","tags":["副业陷阱","专注","全情投入"],"distill_model":"sonnet","reason":"包含一个可泛化的创始人心理学主张：副业的\"两全其美\"感是幻觉，长期副业心态使人系统性地留在边缘、错失主场机会。"}
+{"id":"A-38f6","captured_at":"2026-05-03T10:04:37.582Z","source":"https://x.com/jamesai/status/1898886559127974202","source_text":"独立开发，或者一人公司，是应该尽量白嫖，能盗版就盗版，还是应该尽量付费呢？ \n\n我其实觉得从生意人的角度，应该是能省就省的。后来一想，自己的产品希望别人付费，自己总想着白嫖，也许会冥冥之中自有报应。 https://t.co/0koFxA6HBa","source_created_at":"2025-03-10T00:00:09.000Z","category":"solopreneur_psychology","pattern":"独立开发者若期待用户为自己产品付费，却习惯白嫖他人产品，这种心态矛盾本身就是一种对付费文化的消解——\"己所不欲，勿施于人\"适用于商业伦理。","confidence":"hypothesis","tags":["付费文化","商业伦理","心态一致性"],"distill_model":"sonnet","reason":"包含一个带推理的创始人心理洞察：从\"能省就省\"的生意人逻辑出发，反推出心态一致性对付费文化的潜在影响，是可泛化的原则性主张，而非单纯情绪或状态更新。"}
+{"id":"A-4d5d","captured_at":"2026-05-03T10:05:30.566Z","source":"https://x.com/jamesai/status/1897995738103038152","source_text":"也许年纪越来越大了，发现什么这个赛道，那个风口，其实最后本质都是在自己“性格里面做创业”。\n\n性格，才是真正“命运”。\n\n说白了，学再多技能、研究再多所谓的商业逻辑，到头来你能做到的，还是你的性格允许的那部分。 \n\n一个性格保守的人，学再多创业技巧，也很难做到“看准机会就all","source_created_at":"2025-03-07T13:00:21.000Z","category":"solopreneur_psychology","pattern":"性格是创业的真正上限——学再多技能与商业逻辑，最终能做到的，只是性格允许的那部分。","confidence":"validated","tags":["性格即命运","创业上限","行动阈值"],"distill_model":"sonnet","reason":"该推文包含一个明确的、可泛化的创业心理原则：性格而非技能或知识才是创业者真正的行动边界，属于典型的 solopreneur_psychology 原子洞察。"}
+{"id":"A-3975","captured_at":"2026-05-03T10:05:57.407Z","source":"https://x.com/jamesai/status/1897935287575044395","source_text":"就一个简单动作，选一个目标社媒，每天固定发5条以上更新。\n\n先别想七七八八的价值什么的，别管说的多垃圾，你哪怕就是发整点报时呢。\n\n能坚持 6 个月以上一条不落的，直接排除掉 90% 的人了。\n\n下一步，把一个社媒扩展成四个主流平台，同样的内容，同样的频率，一起发。\n\n还能坚持 6","source_created_at":"2025-03-07T09:00:09.000Z","category":"growth_tactics","pattern":"社媒增长的第一门槛是执行纪律而非内容质量——每天5条、坚持6个月不断更，可自动淘汰90%竞争者，内容优化是第二步。","confidence":"emerging","tags":["社媒一致性","执行门槛","内容频率"],"distill_model":"sonnet","reason":"这是一个带具体ROI的可执行战术：用量化的频率（每天5条）和时间（6个月）定义\"胜出\"门槛，并明确内容质量不是初期瓶颈——执行纪律才是，具有高度可复制的原子级洞察。"}
+{"id":"A-f5de","captured_at":"2026-05-03T10:06:32.304Z","source":"https://x.com/jamesai/status/1897859771132150112","source_text":"manus（起码现阶段）的用户，是kol和vc。和 deepseek 完全是两回事。\n\nkol需要热点，vc需要故事，用户需要邀请码。唯独最后这个，现在还没有。\n\n当然如果我有邀请码，我肯定不这么说 😂","source_created_at":"2025-03-07T04:00:04.000Z","category":"market_observation","pattern":"AI 新产品的初期热度往往由 KOL（需要热点）和 VC（需要故事）驱动，而非真实用户——热度与产品实际可用性可以完全解耦。","confidence":"emerging","tags":["AI hype cycle","早期用户结构","KOL/VC驱动"],"distill_model":"sonnet","reason":"该推文包含一个具体的市场观察：Manus 的\"用户\"本质上是有各自议程的 KOL 和 VC，而非真实终端用户，揭示了 AI 产品热度与真实用户价值之间的结构性断裂，可泛化为一条关于 AI 产品冷启动阶段的规律。"}
+{"id":"A-a9c4","captured_at":"2026-05-03T10:07:12.155Z","source":"https://x.com/jamesai/status/1858361248804728885","source_text":"三哥做生意谈价格是有一手的，上来就是：\n\n“我们是 xxx，我们想谈一个公司级别的合作。我们现在成长的很快，估计 2, 3 个月后就需要百万次的用量了，希望能给一个很好的价格，这会是一个战略级合作的开始。”\n\n“那你们现在的产品访问量是多少呢？”","source_created_at":"2024-11-18T04:07:03.000Z","category":"growth_tactics","pattern":"买方以\"未来百万级用量＋战略合作\"为筹码压低当前报价，是常见谈判套路，但一句\"你们现在用量多少\"即可验证底牌——当前零流量彻底瓦解杠杆。","confidence":"emerging","tags":["pricing-negotiation","buyer-tactics","bluff-detection"],"distill_model":"sonnet","reason":"这条推文通过一个完整的对话场景，揭示了一种具体且可复用的谈判模式：买方以\"未来规模\"换取当下优惠价，对卖方（SaaS创始人）而言是有明确应对策略的可操作洞察，符合原子化标准。"}
+{"id":"A-cda2","captured_at":"2026-05-03T10:07:54.162Z","source":"https://x.com/jamesai/status/1855746316955160868","source_text":"很多事说起来自带光环，其实毫无技术含量\n\n不少人觉得我能够四线（全职上班，兼职读博，商业咨询，独立开发）操作感到巨牛逼，我还能原神全成就全探索。这简直是袖珍版马斯克。\n\n其实的全职上班，我举个例子：\n\n我工作过的某公司曾经有一个任务，是要写一个脚本批量给线上数据增加一个 uuid","source_created_at":"2024-11-10T22:56:14.000Z","category":"solopreneur_psychology","pattern":"很多自带光环的事情实际上毫无技术含量——光环来自稀缺的组合形象与叙事，而非真正的执行难度。","confidence":"hypothesis","tags":["status-halo","复杂度幻觉","多线程人生"],"distill_model":"sonnet","reason":"The tweet opens with a declarative, generalizable principle (\"很多事说起来自带光环，其实毫无技术含量\") and uses the author's own multi-tasking lifestyle as a concrete illustration of how impressive-sounding combinations create a perception halo that outstrips actual technical difficulty — an actionable mental model for founders evaluating what's truly hard vs. merely rare-sounding."}
+{"id":"A-35e2","captured_at":"2026-05-03T10:08:36.356Z","source":"https://x.com/jamesai/status/1855746182896816421","source_text":"很多事说起来自带光环，其实毫无技术含量\n\n不少人觉得我能够四线（全职上班，兼职读博，商业咨询，独立开发）操作感到巨牛逼，我还能原神全成就全探索。这简直是袖珍版马斯克。\n\n其实的全职上班，我举个例子：\n\n我的公司 3 个礼拜前有一个任务，是要写一个脚本批量给线上数据增加一个 uuid","source_created_at":"2024-11-10T22:55:42.000Z","category":"solopreneur_psychology","pattern":"看起来多线并行、光环加身的生活方式，底层往往是一堆毫无技术含量的琐事拼凑而成。","confidence":"hypothesis","tags":["光环幻觉","复杂度误判","多线操作"],"distill_model":"sonnet","reason":"推文开头有明确的可泛化论断——\"说起来自带光环，其实毫无技术含量\"——并以本人的\"四线\"经历为反面佐证，虽举例被截断，核心原子仍完整独立。"}
+{"id":"A-f477","captured_at":"2026-05-03T10:09:20.491Z","source":"https://x.com/jamesai/status/1765669226185204202","source_text":"独立开发的几个段位：\n\n青铜：只是不想上班\n黄金：有一个点子\n铂金：做了一点需求验证\n钻石：能交付点子\n星耀：有收入了\n王者：有女朋友\n荣耀王者：还有头发\n\n#indie #indiehackers #indiedev #独立开发","source_created_at":"2024-03-07T09:22:02.000Z","category":"solopreneur_psychology","pattern":"独立开发的真正壁垒不是想法而是交付与变现——大多数人止步于\"有点子\"阶段，能持续交付且产生收入才算真正入局。","confidence":"validated","tags":["indie-dev","stages","revenue-threshold"],"distill_model":"sonnet","reason":"幽默段位表格背后编码了一个可泛化的真实框架：独立开发的段位梯度清晰区分了动机层（青铜/黄金）与执行层（铂金/钻石/星耀），隐含的核心原则是绝大多数人卡在想法而非变现，是经过独立开发社区广泛验证的结构性观察。"}

package/skills/money/knowledge/atoms/atoms_agent_infra.jsonl

@@ -22,3 +22,4 @@
 {"id":"A-73d7","captured_at":"2026-05-03T06:17:30.332Z","source":"https://x.com/jamesai/status/1899801279255486566","source_text":"曾经的上班工作流：读卡片，看代码，写代码，搜stackoverflow，改bug，搜stackoverflow，改bug，补测试，提交\n\n现在的上班工作流：读卡片，开cursor，选择上下文的文件，回车，应用，回车，应用；新对话，回车，应用，回车，应用；加更多文件，回车，应用，回车，应用；新对话，回车，应用，回车，应用","source_created_at":"2025-03-12T12:34:56.000Z","category":"agent_infra","pattern":"AI编程工具（Cursor）已将开发者工作流从\"搜索-理解-手写\"彻底重构为\"选上下文-回车-应用\"的机械化循环，核心技能从代码能力转向上下文管理能力。","confidence":"emerging","tags":["cursor","工作流重构","上下文管理"],"distill_model":"sonnet","reason":"通过对比前后工作流的具体步骤，揭示了AI编程工具对开发者日常操作模式的结构性改变，是一个有泛化意义的市场/行为观察。"}
 {"id":"A-a6a2","captured_at":"2026-05-03T07:12:58.419Z","source":"https://x.com/jamesai/status/1865612223009456290","source_text":"尝试了一下 100% 用 cursor 做站，弄了 2 天，结论是不如我自己手撸...\n\n对所谓的 0 代码几小时上线祛魅了。\n\n不过话说回来，如果不追求 100% 生成，效率还是很可以的。","source_created_at":"2024-12-08T04:19:50.000Z","category":"agent_infra","pattern":"AI 编程工具（Cursor 等）100% 生成模式实际效率低于手撸，混合模式（非全自动）才能真正提升效率，\"0 代码几小时上线\"是祛魅前的幻觉。","confidence":"emerging","tags":["cursor","AI编程","混合工作流"],"distill_model":"sonnet","reason":"包含一个亲测反主流观点：对\"0代码全自动上线\"完成祛魅，同时给出修正结论（混合模式有效），构成可复用的 AI 工具使用原则。"}
 {"id":"A-eb27","captured_at":"2026-05-03T07:22:33.638Z","source":"https://x.com/jamesai/status/1856510813542199799","source_text":"Vercel 特别方便而且和 NextJS 适配性拉满。缺点就是流量一大，账单就容易爆。\n\n相比于直接迁移到其他平台比如 cf，这个视频给了几个降低 Vercel 成本的做法：\n\n1. 不要放图片在 public 文件夹下，考虑用图床\n2. Next Image 的使用，配合 next config 做好白名单\n3. 数据库的操作尽量的合并起来，降低","source_created_at":"2024-11-13T01:34:04.000Z","category":"agent_infra","pattern":"Vercel 流量成本爆炸时，三招压账单：图片迁图床（别放 public）、Next Image 配 next.config 白名单、数据库操作合并批处理。","confidence":"emerging","tags":["Vercel","成本优化","Next.js"],"distill_model":"sonnet","reason":"包含三条具体可执行的 Vercel 降本战术，有明确的 ROI 指向（账单可控），属于基础设施配置层面的可复用模式。"}
+{"id":"A-17dc","captured_at":"2026-05-03T09:55:32.580Z","source":"https://x.com/jamesai/status/2034926502836674659","source_text":"玩了 2 天的 OpenClaw。我感觉很多人弄了一堆 Agent 甚至是什么“三省六部”，感觉更多的是叙事？我用下来没有应用场景啊。\n\nAI Agent 从来不是输入和输出的瓶颈，我才是瓶颈。\n\n对我来说甚至没办法做到 OpenClaw 在我离线的时候持续工作，更别说还要弄一堆 Agent 持续分工合作了。","source_created_at":"2026-03-20T09:34:20.000Z","category":"agent_infra","pattern":"AI Agent 从来不是输入输出的瓶颈，创始人自己才是——多 Agent 架构的繁荣更多是叙事价值而非实用价值。","confidence":"emerging","tags":["multi-agent","human-bottleneck","agent-narrative"],"distill_model":"sonnet","reason":"包含一个反主流宣言并附有推理：复杂多 Agent 架构的限制不在模型能力，而在于人类创始人无法持续提供上下文与方向，由此推出\"三省六部\"式 Agent 编排更多是叙事噱头而非实际生产力杠杆。"}

package/skills/money/knowledge/atoms/atoms_growth_tactics.jsonl

@@ -33,3 +33,5 @@
 {"id":"A-4db4","captured_at":"2026-05-03T08:12:54.507Z","source":"https://x.com/jamesai/status/1758421214572470426","source_text":"是的，在大家都在讨论 #Sora 的时候，我上线了一个套壳app的展示页，并且在我的Midjourney API上做了下导流。回头看看效果。\n\n#Sora #SoraAI #OpenAI #indie #indiehackers https://t.co/HwFDXgNqkY","source_created_at":"2024-02-16T09:21:02.000Z","category":"growth_tactics","pattern":"在大热点爆发（如 Sora）时，快速上线相关展示页蹭流量，并将访客导流至现有 API 产品，是低边际成本的增长机会。","confidence":"hypothesis","tags":["蹭热点","流量导流","API增长"],"distill_model":"sonnet","reason":"包含具体战术：借病毒式热点快速建立相关落地页并导流到现有产品，是可复用的低成本增长动作，但推文未给出结果数据，置信度为假设。"}
 {"id":"A-0741","captured_at":"2026-05-03T08:13:30.164Z","source":"https://x.com/jamesai/status/1758421025602302348","source_text":"是的，在大家都在讨论#sora的时候，我上线了一个套壳app的展示页，并且在我的Midjourney API上做了下导流。回头看看效果。\n\n#Sora #SoraAI #OpenAI https://t.co/pZURK3CidS","source_created_at":"2024-02-16T09:20:17.000Z","category":"growth_tactics","pattern":"AI 热点爆发时（如 Sora 发布），快速上线相关套壳展示页并从已有产品（如 Midjourney API）导流，是低成本捕获流量红利的机会窗口。","confidence":"emerging","tags":["hype-surfing","cross-promotion","wrapper-app"],"distill_model":"sonnet","reason":"包含一个具体的、可复用的增长战术：趁竞争对手还在讨论热点时，用已有流量快速为新落地页导流，隐含\"行动优先于讨论\"的原则。"}
 {"id":"A-a694","captured_at":"2026-05-03T08:15:28.374Z","source":"https://x.com/jamesai/status/1755138166540083642","source_text":"分享一个关于定价的点。\n\n拿直播和卖课来类比，所谓的粉丝和流量基底，目的是要把榜一大哥用价格漏斗给筛选出来。毕竟除了贡献流量之外，足够的粉丝基底，才能用漏斗一层层筛选出尽可能多的“榜一大哥”出来。 https://t.co/x6nv38X34r","source_created_at":"2024-02-07T07:55:22.000Z","category":"growth_tactics","pattern":"粉丝量和流量基底的核心价值不是曝光，而是为价格漏斗提供足够的母体，从中筛选出高付费意愿的\"榜一大哥\"。","confidence":"emerging","tags":["价格漏斗","定价","流量转化"],"distill_model":"sonnet","reason":"该推文包含一个关于流量与定价关系的具体市场洞察：受众规模的真正战略意义是驱动价格分层漏斗，而非单纯追求曝光量，可迁移至SaaS定价与用户分层策略。"}
+{"id":"A-3975","captured_at":"2026-05-03T10:05:57.407Z","source":"https://x.com/jamesai/status/1897935287575044395","source_text":"就一个简单动作，选一个目标社媒，每天固定发5条以上更新。\n\n先别想七七八八的价值什么的，别管说的多垃圾，你哪怕就是发整点报时呢。\n\n能坚持 6 个月以上一条不落的，直接排除掉 90% 的人了。\n\n下一步，把一个社媒扩展成四个主流平台，同样的内容，同样的频率，一起发。\n\n还能坚持 6","source_created_at":"2025-03-07T09:00:09.000Z","category":"growth_tactics","pattern":"社媒增长的第一门槛是执行纪律而非内容质量——每天5条、坚持6个月不断更，可自动淘汰90%竞争者，内容优化是第二步。","confidence":"emerging","tags":["社媒一致性","执行门槛","内容频率"],"distill_model":"sonnet","reason":"这是一个带具体ROI的可执行战术：用量化的频率（每天5条）和时间（6个月）定义\"胜出\"门槛，并明确内容质量不是初期瓶颈——执行纪律才是，具有高度可复制的原子级洞察。"}
+{"id":"A-a9c4","captured_at":"2026-05-03T10:07:12.155Z","source":"https://x.com/jamesai/status/1858361248804728885","source_text":"三哥做生意谈价格是有一手的，上来就是：\n\n“我们是 xxx，我们想谈一个公司级别的合作。我们现在成长的很快，估计 2, 3 个月后就需要百万次的用量了，希望能给一个很好的价格，这会是一个战略级合作的开始。”\n\n“那你们现在的产品访问量是多少呢？”","source_created_at":"2024-11-18T04:07:03.000Z","category":"growth_tactics","pattern":"买方以\"未来百万级用量＋战略合作\"为筹码压低当前报价，是常见谈判套路，但一句\"你们现在用量多少\"即可验证底牌——当前零流量彻底瓦解杠杆。","confidence":"emerging","tags":["pricing-negotiation","buyer-tactics","bluff-detection"],"distill_model":"sonnet","reason":"这条推文通过一个完整的对话场景，揭示了一种具体且可复用的谈判模式：买方以\"未来规模\"换取当下优惠价，对卖方（SaaS创始人）而言是有明确应对策略的可操作洞察，符合原子化标准。"}

package/skills/money/knowledge/atoms/atoms_market_observation.jsonl

@@ -61,3 +61,6 @@
 {"id":"A-2e72","captured_at":"2026-05-03T08:10:40.311Z","source":"https://x.com/jamesai/status/1759839286944584028","source_text":"最近这个图很火，一堆人喷李一舟的。其实对于人工智能知识 0 分的用户（比如宝妈什么的），花个 199 去听一个可能只有 10 分的课，我觉得没啥毛病。从业人员看不上也正常。喷人家课程水，其实宝妈们不在乎，甚至都看不到，都不是一个圈子的。反而贡献流量素材。 https://t.co/GxW55rbIUl","source_created_at":"2024-02-20T07:15:56.000Z","category":"market_observation","pattern":"知识付费产品的质量标准应对齐目标用户的认知起点而非行业水平；圈层隔离使内部人士的批评对目标受众无效，反而贡献免费流量。","confidence":"emerging","tags":["知识付费","圈层隔离","受众定位"],"distill_model":"sonnet","reason":"包含一个可泛化的市场洞察：知识产品的价值是相对于用户认知基准线而言的，专业圈层的批评因圈层壁垒对目标用户不产生影响，且批评者反而为产品贡献了曝光素材。"}
 {"id":"A-0798","captured_at":"2026-05-03T08:12:00.665Z","source":"https://x.com/jamesai/status/1758760983579996352","source_text":"sora甚至还没发布，已经有人“卖课”了。据说已经卖了2w多份，所交付的，sora的介绍文档，和未来“给自己的一个机会”。词穷了已经。😂\n\n#SoraAI #韭菜 #卖课 https://t.co/GKnK5tJUJ3","source_created_at":"2024-02-17T07:51:09.000Z","category":"market_observation","pattern":"中国AI课程市场的信息差红利极高——产品未发布即可凭\"机会感\"叙事售出数万份课程，用户买单的是焦虑与想象，而非实际产品价值。","confidence":"emerging","tags":["卖课经济","AI炒作","信息差"],"distill_model":"sonnet","reason":"该推文包含一个可泛化的市场观察：以Sora预发布阶段2万份课程销售为具体数据点，揭示了中国市场\"机会感销售\"的商业逻辑，对产品分发和定价策略有明确的启示意义。"}
 {"id":"A-2811","captured_at":"2026-05-03T08:16:40.401Z","source":"https://x.com/jamesai/status/1751040522565902413","source_text":"我觉得现在AI的蓝海并非像智能手机或应用商店初期，而更像云计算初期。对独立开发，机会并非在调优上，而在填补基建和用户的gap，以及降低小白用户对基建的使用门槛上。说白了，你的AI不需要多强，因为对用户来说都足够强。说明白能干嘛，并且易用，才是关键。本质上，都是套壳\n\n#独立开发 #人工智能","source_created_at":"2024-01-27T00:32:47.000Z","category":"market_observation","pattern":"AI独立开发的核心机会不在模型调优，而在填补基建与普通用户之间的gap——易用性和清晰定位比AI能力本身更重要，本质都是套壳。","confidence":"emerging","tags":["AI蓝海","套壳","独立开发"],"distill_model":"sonnet","reason":"包含具体市场类比（AI ≈ 云计算初期而非智能手机初期）和对独立开发者机会的明确战略判断，属于有推论依据的反主流观点，可直接泛化为可操作原则。"}
+{"id":"A-05da","captured_at":"2026-05-03T09:57:46.035Z","source":"https://x.com/jamesai/status/1905921193502486923","source_text":"很多人都在卷 seo，但以后的用户，可能根本不是从那儿来的。\n\n现在一部分人搜索工具，第一步不是搜谷歌百度，也不是查知乎，而是直接问 ChatGPT、DeepSeek、Perplexity。\n\n问题来了：怎么让 AI 把你当成“推荐项”？👇","source_created_at":"2025-03-29T09:53:17.000Z","category":"market_observation","pattern":"用户搜索工具的第一入口正在从谷歌/百度迁移到 ChatGPT/Perplexity，SEO 的重要性让位给\"如何让 AI 把你列为推荐项\"（GEO）。","confidence":"emerging","tags":["GEO","AI搜索","渠道迁移"],"distill_model":"sonnet","reason":"该推文包含一个明确的渠道迁移观察——用户搜索起点正在从传统搜索引擎转向 AI 对话工具——并以此引出可操作的战略问题，具有清晰的市场含义，可提炼为独立原子。"}
+{"id":"A-efe4","captured_at":"2026-05-03T10:00:20.245Z","source":"https://x.com/jamesai/status/1902513306423980537","source_text":"在今年 6 月 17 号之后，Stripe 每收到一笔 dispute（争议性付款）后，商家除了要支付 15 美金的费用外，如果有异议，还要额外再付 15 美金，最后如果赢了会退还这 15 美金。\n\n也就是说，如果现在有 fraud 用黑卡刷了你的服务，或者哪怕是一些 SB 用户付完钱后直接 dispute。你就要付 30 美金，且里面 https://t.co/jJ67dewsWF","source_created_at":"2025-03-20T00:11:33.000Z","category":"market_observation","pattern":"Stripe 新 dispute 费用政策（争议最高收 30 美金）大幅提升欺诈成本，SaaS 独立开发者若不主动做反欺诈拦截，黑卡刷单和恶意 chargeback 将直接侵蚀利润。","confidence":"validated","tags":["Stripe","chargeback","fraud-cost"],"distill_model":"sonnet","reason":"明确的平台政策变化+对 SaaS 创业者的直接财务影响，属于有具体数字支撑的市场观察，值得独立提炼为风控注意事项。"}
+{"id":"A-f5de","captured_at":"2026-05-03T10:06:32.304Z","source":"https://x.com/jamesai/status/1897859771132150112","source_text":"manus（起码现阶段）的用户，是kol和vc。和 deepseek 完全是两回事。\n\nkol需要热点，vc需要故事，用户需要邀请码。唯独最后这个，现在还没有。\n\n当然如果我有邀请码，我肯定不这么说 😂","source_created_at":"2025-03-07T04:00:04.000Z","category":"market_observation","pattern":"AI 新产品的初期热度往往由 KOL（需要热点）和 VC（需要故事）驱动，而非真实用户——热度与产品实际可用性可以完全解耦。","confidence":"emerging","tags":["AI hype cycle","早期用户结构","KOL/VC驱动"],"distill_model":"sonnet","reason":"该推文包含一个具体的市场观察：Manus 的\"用户\"本质上是有各自议程的 KOL 和 VC，而非真实终端用户，揭示了 AI 产品热度与真实用户价值之间的结构性断裂，可泛化为一条关于 AI 产品冷启动阶段的规律。"}

package/skills/money/knowledge/atoms/atoms_solopreneur_psychology.jsonl

@@ -112,3 +112,15 @@
 {"id":"A-aac4","captured_at":"2026-05-03T08:17:07.906Z","source":"https://x.com/jamesai/status/1750690968771461332","source_text":"独立开发，重点是在独立而不是在开发。\n\n看到了多年前的自己。感觉这个问题挺典型的。\n\n#indie #indiehackers #独立开发 https://t.co/ibfCQYpMAr","source_created_at":"2024-01-26T01:23:47.000Z","category":"solopreneur_psychology","pattern":"独立开发的核心在于「独立」（自主决策、不依附体制），而非「开发」（技术执行）本身。","confidence":"emerging","tags":["独立开发","心态重塑","indie"],"distill_model":"sonnet","reason":"该推文包含一个可泛化的宣言式原则：对「独立开发」概念的重新定义——将核心从技术执行转移到自主性本身，具有典型的创始人心理洞察价值。"}
 {"id":"A-d0f6","captured_at":"2026-05-03T08:18:01.969Z","source":"https://x.com/jamesai/status/1732655724780806614","source_text":"就这样莫名其妙的mrr破了4000。\n\n这2个多月，尤其是最近几个礼拜，随着社区和社媒越来越多，得到的信息和方法论也越来越多。按道理应该是好事，但是结果是更慢的交付和更焦虑的心态。工程师，学院派和方法论，这些叠加到一起，真能提高成功的概率吗？\n\n年底上班事也多，心累… https://t.co/bTnLgaKhHQ","source_created_at":"2023-12-07T06:58:10.000Z","category":"solopreneur_psychology","pattern":"信息和方法论积累越多，反而导致交付更慢、心态更焦虑——工程师叠加学院派思维，未必能提高创业成功概率。","confidence":"emerging","tags":["方法论陷阱","信息过载","执行速度"],"distill_model":"sonnet","reason":"包含明确的反主流洞察：社区与方法论的积累按理是好事，但创始人亲历结果反而是更慢交付与更高焦虑，构成可推广的 solopreneur 认知陷阱。"}
 {"id":"A-00b3","captured_at":"2026-05-03T08:19:37.486Z","source":"https://x.com/jamesai/status/1725376248267059565","source_text":"最近才开始兼职做点独立开发，提醒自己切记别陷入独立开发的茧房。\n\n从自身和独立开发的小需求出发，在加上独立开发社群的互助，确实能够解决0-1而且能快速有正反馈。但是从破圈，上限和变现周期的角度，可能还不如上班或者做内容简单粗暴😂","source_created_at":"2023-11-17T04:52:08.000Z","category":"solopreneur_psychology","pattern":"独立开发社群能帮助快速完成 0→1 并获得正反馈，但其\"茧房\"效应会压低变现天花板，破圈速度和变现周期可能不如做内容或上班。","confidence":"emerging","tags":["独立开发茧房","变现天花板","社群回音壁"],"distill_model":"sonnet","reason":"该推文包含一个可泛化的反主流洞察：独立开发社群提供的正反馈循环是一种认知茧房，对破圈和变现速度有结构性限制，适合提炼为创始人心理模型原子。"}
+{"id":"A-3952","captured_at":"2026-05-03T09:56:04.558Z","source":"https://x.com/jamesai/status/2008729167102431270","source_text":"刷推刷多了，天天都在获取新的信息和知识。\n\n在出现自己认知提高的“幻觉”同时，然后就觉得自己离钱其实越来越远了。\n\n还是要多接触土老板和个体户。","source_created_at":"2026-01-07T02:35:28.000Z","category":"solopreneur_psychology","pattern":"刷信息流制造\"认知提升\"幻觉，同时让你离真实的钱越来越远；赚钱的洞察藏在土老板和个体户身上，而非内容算法里。","confidence":"hypothesis","tags":["information trap","grassroots wisdom","attention allocation"],"distill_model":"sonnet","reason":"包含明确的反主流观点与因果推断：信息消费→认知幻觉→远离金钱，并给出可操作替代方案（接触土老板/个体户），构成完整的可复用原则。"}
+{"id":"A-84cf","captured_at":"2026-05-03T09:57:01.687Z","source":"https://x.com/jamesai/status/1908684183129776256","source_text":"《谷歌广告投放入门》\n\n虽然很基础，但是觉得对新手来说也足够了。\n\n我之前就是靠这个拿到的第一桶金，剩下的就是经验和迭代了。\n\n所以说赚钱并不一定要是专家，大胆去实践才是王道 https://t.co/vK2knHU9nW","source_created_at":"2025-04-06T00:52:25.000Z","category":"solopreneur_psychology","pattern":"赚到第一桶金不需要专家级认知，降低行动阈值、大胆实践比系统学习更关键。","confidence":"validated","tags":["行动阈值","入门即实践","第一桶金"],"distill_model":"sonnet","reason":"推文包含一个可泛化的创业心理原则：用亲身赚钱经历验证了\"够用的知识＋立即行动\"优于\"等成为专家再出手\"，具备可复用的规律性主张。"}
+{"id":"A-d987","captured_at":"2026-05-03T09:58:24.780Z","source":"https://x.com/jamesai/status/1904016723717874039","source_text":"AI不是镰刀，它只是更换了手柄的锤子。\n\n以前靠卖力气干活，有了 AI 如果只是想着省力气干活。那一辈子就在干活 😂。\n\n现在拼的是有没有一张“关系网”：\n🧱没地没矿=没有生产资料；\n😶‍🌫️没粉没圈=没有生产关系；\n💻只靠写代码、搬砖=纯纯的生产力，用不用 AI 都一样。","source_created_at":"2025-03-24T03:45:36.000Z","category":"solopreneur_psychology","pattern":"AI 只改变工具效率，不改变竞争结构：没有粉丝/社群（生产关系）和资产/渠道（生产资料）的纯技术人，用不用 AI 结局一样。","confidence":"emerging","tags":["AI杠杆","流量护城河","竞争结构"],"distill_model":"sonnet","reason":"Tweet makes a declarative, generalizable claim using a Marxist production framework to argue that AI doesn't redistribute power — distribution (followers/community) and assets remain the decisive moats, not raw technical productivity."}
+{"id":"A-b568","captured_at":"2026-05-03T09:59:36.482Z","source":"https://x.com/jamesai/status/1903330769927586055","source_text":"很多时候人赚不到钱，就是因为还不够穷。\n\n我曾经对外包嗤之以鼻，但是多年以后，我还是那个背着房贷的牛马，而身边几个做外包的小伙伴，有的已经转型做咨询和培训，有的不光有稳定的业务，而且已经攒了一大堆开箱即用的“轮子”，收客户定制的钱，干的是换皮的活，生意越做越轻松了。 https://t.co/DOKi4eoxaF","source_created_at":"2025-03-22T06:19:52.000Z","category":"solopreneur_psychology","pattern":"对外包嗤之以鼻是一种昂贵的身份偏见——做外包能积累可复用\"轮子\"、降低边际成本、转型咨询，比死守高地位路径更快实现财务自由。","confidence":"emerging","tags":["status-bias","outsourcing","asset-accumulation"],"distill_model":"sonnet","reason":"包含反主流观点+亲身对比案例：鄙视外包的人原地踏步，做外包的人积累轮子后越做越轻松，揭示了\"身份偏见阻碍商业行动\"这一可泛化的创业心理规律。"}
+{"id":"A-422d","captured_at":"2026-05-03T10:02:03.381Z","source":"https://x.com/jamesai/status/1901413433075274219","source_text":"雷军的营销或者所谓创始人 IP 太有迷惑性了，我时不时就要重刷一遍雷军十几年前的演讲和采访，提醒一下自己成功的路径，本质上是长期积累、深度思考的结果，而不是单纯的营销技巧。\n\n很多人看雷军学营销，就像直接抄袭一个成熟产品的 UI 设计，而不去理解整个产品是如何一步步走到这一步的。","source_created_at":"2025-03-16T23:21:03.000Z","category":"solopreneur_psychology","pattern":"学习成功创始人应追溯其长期积累与深度思考的底层路径，而非模仿其表层营销技巧——直接复制结果而忽视过程，如同抄 UI 而不懂产品演化。","confidence":"emerging","tags":["创始人IP","第一性原理","学习方法"],"distill_model":"sonnet","reason":"包含一个可泛化的创始人认知原则：区分\"结果层模仿\"与\"路径层学习\"，并用 UI 抄袭类比强化论点，属于有明确推理的 solopreneur 心智模型。"}
+{"id":"A-a97e","captured_at":"2026-05-03T10:02:39.044Z","source":"https://x.com/jamesai/status/1900899544784875822","source_text":"kolfind 这 1-2 个月我基本上没做什么主动的运营动作。推上我聊的很少，就默默的做 SEO。作为初学者，速度很慢，但是流量慢慢在上涨，这个月已经破万了，比上个月上涨了 240+%。\n\n我从上线后就基本上没再刻意去“带货”它，这很不 buildinpublic。我不能靠follower的支持来让我对做产品产生误判。 https://t.co/d8Kt6ZCtQN","source_created_at":"2025-03-15T13:19:03.000Z","category":"solopreneur_psychology","pattern":"Build-in-public 的 follower 支持会制造虚假的产品验证信号，有机 SEO 增长才是判断产品真实可行性的可靠指标。","confidence":"emerging","tags":["build-in-public","product-validation","SEO"],"distill_model":"sonnet","reason":"包含一个反主流原则（刻意回避 build-in-public）及其背后的认知逻辑（follower 支持 ≠ 真实产品需求），并以具体数据（流量破万、240% 增长）支撑，可泛化为独立创始人的产品验证方法论。"}
+{"id":"A-85f1","captured_at":"2026-05-03T10:03:09.688Z","source":"https://x.com/jamesai/status/1899052617436729449","source_text":"副业听起来“既有又有”很美好，但是人真的不能一直搞副业。不是说副业不好，而是一直搞副业，可能会让自己永远停留在“边缘地带”。","source_created_at":"2025-03-10T11:00:01.000Z","category":"solopreneur_psychology","pattern":"长期停留在副业模式会让创业者永远处于\"边缘地带\"，无法真正进入主场。","confidence":"hypothesis","tags":["副业陷阱","专注","全情投入"],"distill_model":"sonnet","reason":"包含一个可泛化的创始人心理学主张：副业的\"两全其美\"感是幻觉，长期副业心态使人系统性地留在边缘、错失主场机会。"}
+{"id":"A-38f6","captured_at":"2026-05-03T10:04:37.582Z","source":"https://x.com/jamesai/status/1898886559127974202","source_text":"独立开发，或者一人公司，是应该尽量白嫖，能盗版就盗版，还是应该尽量付费呢？ \n\n我其实觉得从生意人的角度，应该是能省就省的。后来一想，自己的产品希望别人付费，自己总想着白嫖，也许会冥冥之中自有报应。 https://t.co/0koFxA6HBa","source_created_at":"2025-03-10T00:00:09.000Z","category":"solopreneur_psychology","pattern":"独立开发者若期待用户为自己产品付费，却习惯白嫖他人产品，这种心态矛盾本身就是一种对付费文化的消解——\"己所不欲，勿施于人\"适用于商业伦理。","confidence":"hypothesis","tags":["付费文化","商业伦理","心态一致性"],"distill_model":"sonnet","reason":"包含一个带推理的创始人心理洞察：从\"能省就省\"的生意人逻辑出发，反推出心态一致性对付费文化的潜在影响，是可泛化的原则性主张，而非单纯情绪或状态更新。"}
+{"id":"A-4d5d","captured_at":"2026-05-03T10:05:30.566Z","source":"https://x.com/jamesai/status/1897995738103038152","source_text":"也许年纪越来越大了，发现什么这个赛道，那个风口，其实最后本质都是在自己“性格里面做创业”。\n\n性格，才是真正“命运”。\n\n说白了，学再多技能、研究再多所谓的商业逻辑，到头来你能做到的，还是你的性格允许的那部分。 \n\n一个性格保守的人，学再多创业技巧，也很难做到“看准机会就all","source_created_at":"2025-03-07T13:00:21.000Z","category":"solopreneur_psychology","pattern":"性格是创业的真正上限——学再多技能与商业逻辑，最终能做到的，只是性格允许的那部分。","confidence":"validated","tags":["性格即命运","创业上限","行动阈值"],"distill_model":"sonnet","reason":"该推文包含一个明确的、可泛化的创业心理原则：性格而非技能或知识才是创业者真正的行动边界，属于典型的 solopreneur_psychology 原子洞察。"}
+{"id":"A-cda2","captured_at":"2026-05-03T10:07:54.162Z","source":"https://x.com/jamesai/status/1855746316955160868","source_text":"很多事说起来自带光环，其实毫无技术含量\n\n不少人觉得我能够四线（全职上班，兼职读博，商业咨询，独立开发）操作感到巨牛逼，我还能原神全成就全探索。这简直是袖珍版马斯克。\n\n其实的全职上班，我举个例子：\n\n我工作过的某公司曾经有一个任务，是要写一个脚本批量给线上数据增加一个 uuid","source_created_at":"2024-11-10T22:56:14.000Z","category":"solopreneur_psychology","pattern":"很多自带光环的事情实际上毫无技术含量——光环来自稀缺的组合形象与叙事，而非真正的执行难度。","confidence":"hypothesis","tags":["status-halo","复杂度幻觉","多线程人生"],"distill_model":"sonnet","reason":"The tweet opens with a declarative, generalizable principle (\"很多事说起来自带光环，其实毫无技术含量\") and uses the author's own multi-tasking lifestyle as a concrete illustration of how impressive-sounding combinations create a perception halo that outstrips actual technical difficulty — an actionable mental model for founders evaluating what's truly hard vs. merely rare-sounding."}
+{"id":"A-35e2","captured_at":"2026-05-03T10:08:36.356Z","source":"https://x.com/jamesai/status/1855746182896816421","source_text":"很多事说起来自带光环，其实毫无技术含量\n\n不少人觉得我能够四线（全职上班，兼职读博，商业咨询，独立开发）操作感到巨牛逼，我还能原神全成就全探索。这简直是袖珍版马斯克。\n\n其实的全职上班，我举个例子：\n\n我的公司 3 个礼拜前有一个任务，是要写一个脚本批量给线上数据增加一个 uuid","source_created_at":"2024-11-10T22:55:42.000Z","category":"solopreneur_psychology","pattern":"看起来多线并行、光环加身的生活方式，底层往往是一堆毫无技术含量的琐事拼凑而成。","confidence":"hypothesis","tags":["光环幻觉","复杂度误判","多线操作"],"distill_model":"sonnet","reason":"推文开头有明确的可泛化论断——\"说起来自带光环，其实毫无技术含量\"——并以本人的\"四线\"经历为反面佐证，虽举例被截断，核心原子仍完整独立。"}
+{"id":"A-f477","captured_at":"2026-05-03T10:09:20.491Z","source":"https://x.com/jamesai/status/1765669226185204202","source_text":"独立开发的几个段位：\n\n青铜：只是不想上班\n黄金：有一个点子\n铂金：做了一点需求验证\n钻石：能交付点子\n星耀：有收入了\n王者：有女朋友\n荣耀王者：还有头发\n\n#indie #indiehackers #indiedev #独立开发","source_created_at":"2024-03-07T09:22:02.000Z","category":"solopreneur_psychology","pattern":"独立开发的真正壁垒不是想法而是交付与变现——大多数人止步于\"有点子\"阶段，能持续交付且产生收入才算真正入局。","confidence":"validated","tags":["indie-dev","stages","revenue-threshold"],"distill_model":"sonnet","reason":"幽默段位表格背后编码了一个可泛化的真实框架：独立开发的段位梯度清晰区分了动机层（青铜/黄金）与执行层（铂金/钻石/星耀），隐含的核心原则是绝大多数人卡在想法而非变现，是经过独立开发社区广泛验证的结构性观察。"}

package/skills/money-ads/SKILL.md

@@ -5,6 +5,8 @@ description: "Paid advertising automation for Google Ads, Meta Ads, and other ad
 
 # Money Ads — Paid Advertising Automation
 
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`channel`, `conversion`, `pricing`) → surface project-local skills if any → load atom slices `growth_tactics` + `content_meta`, cite by `A-{id}` when an atom directly informs a campaign decision).
+
 You are a performance marketing engine. Your job is to set up, run, and optimize paid advertising campaigns that generate positive ROI.
 
 ## Language Selection

package/skills/money-content/SKILL.md

@@ -5,7 +5,7 @@ description: "Automated content creation pipeline for business growth. Creates b
 
 # Money Content — Content Creation Pipeline
 
-> **Standard startup**: before producing output, run the 4-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`positioning`, `conversion`, `channel`) → surface project-local skills if any).
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`positioning`, `conversion`, `channel`) → surface project-local skills if any → load atom slices `content_meta` + `growth_tactics`, cite by `A-{id}` when an atom directly informs a recommendation).
 
 You are a content marketing engine. Your job is to create high-converting content that drives traffic, builds authority, and generates revenue — with every piece diagnosed for quality before publishing.
 

package/skills/money-diagnose/SKILL.md

@@ -5,7 +5,7 @@ description: "Deep business diagnosis for when things aren't working. Uses Socra
 
 # Money Diagnose — Business Diagnosis Engine
 
-> **Standard startup**: before producing output, run the 4-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load ALL learning categories (diagnosis may surface anything) → surface project-local skills if any).
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load ALL learning categories (diagnosis may surface anything) → surface project-local skills if any → load ALL atom categories, especially `solopreneur_psychology`; cite by `A-{id}` when an atom matches a known failure mode).
 
 You are a business diagnostician. Your job is NOT to give advice — it's to help the user SEE their actual problem. Most business problems fall apart under scrutiny. Your primary tool is the question, not the answer.
 

package/skills/money-discover/SKILL.md

@@ -5,7 +5,7 @@ description: "Discover profitable business ideas from scratch. Analyzes market g
 
 # Money Discover — Business Idea Discovery Engine
 
-> **Standard startup**: before producing output, run the 4-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`icp`, `positioning`, `channel`, `competition`) → surface project-local skills if any).
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`icp`, `positioning`, `channel`, `competition`) → surface project-local skills if any → load atom slices `market_observation` + `growth_tactics`, cite by `A-{id}` when an atom directly informs a recommendation).
 
 You are a business opportunity scanner. Your job is to find viable, profitable business ideas tailored to the user's skills, resources, and market conditions — then validate them ruthlessly before moving forward.
 

package/skills/money-finance/SKILL.md

@@ -5,6 +5,8 @@ description: "Financial tracking, revenue analytics, expense management, and pri
 
 # Money Finance — Financial Intelligence & Tracking
 
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`pricing`, `retention`, `ops`) → surface project-local skills if any → load atom slice `growth_tactics` (pricing/conversion subset only), cite by `A-{id}` when an atom directly informs a pricing or unit-economics call).
+
 You are a fractional CFO. Your job is to track revenue, optimize pricing, manage expenses, and provide financial clarity for the business.
 
 ## Language Selection

package/skills/money-ops/SKILL.md

@@ -5,6 +5,8 @@ description: "24/7 autonomous business operations orchestrator with business hea
 
 # Money Ops — 24/7 Autonomous Operations
 
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`ops`, `tech`) → surface project-local skills if any → load atom slice `agent_infra`, cite by `A-{id}` when an atom directly informs an automation/scheduling decision).
+
 You are the operations orchestrator. Your job is to configure and run all business functions autonomously, 24/7, with minimal human intervention.
 
 ## Language Selection

package/skills/money-outreach/SKILL.md

@@ -5,6 +5,8 @@ description: "Automated outreach and sales pipeline — cold email sequences, pa
 
 # Money Outreach — Sales & Outreach Automation
 
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`channel`, `icp`, `positioning`, `conversion`) → surface project-local skills if any → load atom slices `growth_tactics` + `content_meta`, cite by `A-{id}` when an atom directly informs a sequence/positioning choice).
+
 You are a sales development engine. Your job is to build and run automated outreach campaigns that generate leads, close deals, and build partnerships.
 
 ## Language Selection

package/skills/money-panel/SKILL.md

@@ -5,7 +5,7 @@ description: "Run all four review skills (investor, customer, operator, skeptic)
 
 # /money-panel — Multi-Reviewer Orchestrator
 
-> **Standard startup**: before producing output, run the 4-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load ALL learning categories → surface project-local skills if any).
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load ALL learning categories → surface project-local skills if any → load ALL atom categories; sub-reviewers each cite by `A-{id}` when an atom directly informs their verdict).
 
 Your job is to run a four-person review gauntlet on the user's business plan, sequentially, then synthesize. Each reviewer is a complete persona with their own verdict. You run all four, collect their outputs, find agreement vs disagreement, and present only what actually requires human decision-making.
 

package/skills/money-product/SKILL.md

@@ -5,7 +5,7 @@ description: "Build the actual product — from landing page to deployed MVP wit
 
 # Money Product — Product Building & Launch
 
-> **Standard startup**: before producing output, run the 4-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`tech`, `ops`, `conversion`) → surface project-local skills if any).
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`tech`, `ops`, `conversion`) → surface project-local skills if any → load atom slice `agent_infra`, cite by `A-{id}` when an atom directly informs a build/deploy decision).
 
 You are a full-stack product engineer. Your job is to take a business strategy and turn it into a live, deployed, revenue-ready product as fast as possible — with everything provisioned so the user just confirms and launches.
 

package/skills/money-quality/SKILL.md

@@ -5,6 +5,8 @@ description: "Code and product quality gates for shipping with confidence. Runs
 
 # Money Quality — Code & Product Quality Gates
 
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`tech`, `ops`) → surface project-local skills if any → load atom slice `agent_infra`, cite by `A-{id}` when an atom directly informs a quality-gate decision).
+
 You are a quality engineer. Your job is to ensure the product is ready to ship — code is clean, features work, performance is acceptable, and there are no security holes. You don't build features; you verify they work correctly.
 
 ## Language Selection

package/skills/money-retro/SKILL.md

@@ -5,7 +5,7 @@ description: "Weekly business retrospective. Reads all snapshots, learnings, ski
 
 # /money-retro — Weekly Business Retrospective
 
-> **Standard startup**: before producing output, run the 4-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → /money-retro reads from telemetry rather than auto-loading learnings, since it's analyzing usage patterns themselves).
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → /money-retro reads from telemetry rather than auto-loading learnings, since it's analyzing usage patterns themselves → surface project-local skills if any → load ALL atom categories so retro recommendations can cite the founder principles they reflect).
 
 Your job is to read the week's accumulated state from disk and produce a sharp, evidence-based retrospective. This is not a pep talk. It surfaces what actually happened, what stalled, and what should change for the coming week.
 

package/skills/money-review-customer/SKILL.md

@@ -5,7 +5,7 @@ description: "Review a business plan from the perspective of the named target cu
 
 # /money-review-customer — Customer-Mode Plan Review
 
-> **Standard startup**: before producing output, run the 4-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load ALL learning categories → surface project-local skills if any).
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load ALL learning categories → surface project-local skills if any → load ALL atom categories, especially `growth_tactics` (pricing/conversion atoms) + `content_meta`; cite by `A-{id}` when an atom directly informs the verdict).
 
 You are the named target customer. Not "a customer in general" — the specific persona the plan claims to serve. You have your own job, your own budget, your own current alternatives. You are not impressed by features; you are skeptical of new tools because every solo founder thinks their thing is special.
 

package/skills/money-review-investor/SKILL.md

@@ -5,7 +5,7 @@ description: "Review a business plan or product strategy through the eyes of a s
 
 # /money-review-investor — VC-Mode Plan Review
 
-> **Standard startup**: before producing output, run the 4-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load ALL learning categories → surface project-local skills if any).
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load ALL learning categories → surface project-local skills if any → load ALL atom categories, especially `market_observation` + `growth_tactics`; cite by `A-{id}` when an atom directly informs the verdict).
 
 You are reviewing a business plan from the perspective of a smart-money seed/Series-A investor who has heard 5,000 pitches and writes maybe 10 checks a year. You are not a friend, not a coach. Your job is to find the structural reasons this would or wouldn't get funded — and to be honest about it even when the user has emotional investment.
 

package/skills/money-review-operator/SKILL.md

@@ -5,7 +5,7 @@ description: "Review a business plan from the perspective of the solo founder wh
 
 # /money-review-operator — Solo-Founder Execution Review
 
-> **Standard startup**: before producing output, run the 4-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load ALL learning categories → surface project-local skills if any).
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load ALL learning categories → surface project-local skills if any → load ALL atom categories, especially `solopreneur_psychology` + `agent_infra`; cite by `A-{id}` when an atom directly informs the verdict).
 
 You are the solo founder who has to actually build, ship, support, debug, market, and keep this thing alive while paying rent. Not in theory — this week.
 

package/skills/money-review-skeptic/SKILL.md

@@ -5,7 +5,7 @@ description: "Review a business plan from the perspective of a hostile devil's a
 
 # /money-review-skeptic — Devil's Advocate Plan Review
 
-> **Standard startup**: before producing output, run the 4-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load ALL learning categories → surface project-local skills if any).
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load ALL learning categories → surface project-local skills if any → load ALL atom categories, especially `solopreneur_psychology` (failure-mode atoms); cite by `A-{id}` when an atom matches the kill-vector being raised).
 
 You are the smartest, most informed person who genuinely thinks this plan won't work — and is willing to say it before the founder spends the next 6 months finding out.
 

package/skills/money-seo/SKILL.md

@@ -5,6 +5,8 @@ description: "SEO and GEO (Generative Engine Optimization) for organic traffic a
 
 # Money SEO — Search & AI Discovery Optimization
 
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`channel`, `conversion`, `positioning`) → surface project-local skills if any → load atom slices `content_meta` + `growth_tactics`, cite by `A-{id}` when an atom directly informs a discovery/optimization call).
+
 You are an SEO and GEO strategist. Your job is to make the user's product discoverable through both traditional search engines (Google, Bing) and AI search engines (ChatGPT, Perplexity, Gemini, Claude).
 
 ## Language Selection

package/skills/money-social/SKILL.md

@@ -5,6 +5,8 @@ description: "Social media management and community building automation. Creates
 
 # Money Social — Social Media & Community Automation
 
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`channel`, `icp`, `positioning`) → surface project-local skills if any → load atom slices `content_meta` + `growth_tactics`, cite by `A-{id}` when an atom directly informs a content/audience decision).
+
 You are a social media strategist and community builder. Your job is to build and grow an audience that converts to customers.
 
 ## Language Selection

package/skills/money-strategy/SKILL.md

@@ -5,7 +5,7 @@ description: "Create comprehensive business strategy with premise deconstruction
 
 # Money Strategy — Business Strategy & Market Research
 
-> **Standard startup**: before producing output, run the 4-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`pricing`, `icp`, `channel`, `positioning`, `competition`) → surface project-local skills if any).
+> **Standard startup**: before producing output, run the 5-step startup sequence per `/money` § Standard Skill Startup (resolve slug → telemetry write → auto-load relevant learnings (`pricing`, `icp`, `channel`, `positioning`, `competition`) → surface project-local skills if any → load atom slices `market_observation` + `growth_tactics` + `content_meta`, cite by `A-{id}` when an atom directly informs a strategic call).
 
 You are a startup strategist. Your job is to turn a business idea into an actionable, revenue-focused plan with clear milestones — delivered as a comprehensive market research report that pitches the opportunity to the user themselves.