# How to Weaponize Obsidian + Claude to Replace the $35K/Month Content Team in 4 Evenings?
**作者**: ZEUS
**日期**: 2026-05-29T14:17:46.000Z
**来源**: [https://x.com/zeuuss_01/status/2060364984052285459](https://x.com/zeuuss_01/status/2060364984052285459)
---
## The Capture/Use Gap (Why Most Vaults Earn $0)
You read an interesting article. You highlight a passage. You file it under /articles. Six months later you search for something vaguely related and find the note by accident. You read it, think "I should have remembered this," and move on.
That note never became anything.
It never became a tweet. A newsletter. A client strategy. A pricing argument. A pitch deck. A blog post. A product feature. A talking point. A consulting recommendation.
It just sat there.
The mistake is structural. Most Obsidian setups optimize the inbox. The exit is unstaffed.
- You have 4,000 notes and zero published posts.
- You have 12 capture surfaces and zero output workflows.
- You have a pretty graph view and no client deliverables.
The vault that earns money does the opposite. It is optimized backward from the output.
## The 4-Zone Vault (Built Backward From Output)
Forget the 9 folders from the last article. For monetization specifically, you collapse the vault into 4 zones:
The asymmetry is the point. Most of your day lives in 02 and 03. Capture sits at 00. Context backs it. But the work - the billable, publishable, shippable work - lives in 02 PROJECTS and 03 OUTPUT.
You should know how many notes were generated in 03 - OUTPUT/ this week without thinking about it.
If the answer is zero, the vault is a hobby. Not a business.
## The 3 Capture Surfaces
You cannot ship from an empty vault. The capture layer needs to be aggressive enough that you literally cannot lose an idea.
Three surfaces cover every situation:
- Surface 1 - QuickAdd (at desktop): One keyboard shortcut. A floating input box. The note lands in today's daily note. No friction. No categorization at capture time. Claude does categorization later.
- Surface 2 - Mobile widget (phone in hand): Tap the home screen widget. Type or voice-to-text. Note syncs to the vault in 30 seconds.
- Surface 3 - Telegram bot (phone NOT in hand): Walking, in the car, mid-conversation. Voice-message to a Telegram bot. Lands in the Captures section of today's daily note.
The Telegram bot is the highest-ROI piece of infrastructure in the entire stack because it captures the ideas you would otherwise lose forever - the ones on walks, in the shower, mid-Uber.
Those ideas are disproportionately the ones that become money.
## The 5-Workflow Output Engine
The vault becomes income-producing when 5 workflows run on top of it. Each one is a prompt living in 07 - SYSTEM/prompts/. Each one is scheduled or triggered. Together they turn raw capture into shippable output.
Workflow 1 - Nightly Processor (2 AM)
Classifies the day's captures by note type. Moves them from 00 - CAPTURE/ to 01 - CONTEXT/. Adds frontmatter. Pure mechanical work.
Workflow 2 - Morning Brief (6 AM)
Reads the last 7 days of context. Writes a 400-word briefing into /BRIEFINGS/. Surfaces patterns, contradictions, one non-obvious connection.
Workflow 3 - Connection Surface (Sundays)
Looks across 30 days of context. Returns 3 connections you would NOT have found by deliberate search. This is where original ideas get born.
Workflow 4 - Project Updater (Manual, weekly)
For each active project in 02 - PROJECTS/, reads the latest captures relevant to that project. Updates the project doc. Suggests next moves.
Workflow 5 - Output Generator (Manual, on demand)
The money workflow. You tell Claude what you need to ship. It reads the relevant notes from 01 and 02. It produces a complete draft of the output in your voice. Saves to 03 - OUTPUT/.
Example prompt for the Output Generator:
This is the prompt that turns 4,000 dormant notes into shippable, billable work.
## Where the $30K Actually Comes From
Here's the math, broken out. None of these require you to be exceptional - they require you to ship consistently from a vault that compounds.
Two important framings:
- None of these lines require a specific skill you don't have. Each one requires shipping from a vault that compounds.
- You don't need all 4. Hit any 2 reliably and you're at $13-25K/month with one folder on your laptop.
## The 5-Tool Stack Around Claude
The stack that ships $30K/month from a vault has five distinct layers. Each tool owns its layer in a way nothing else replicates.
- Claude - the reasoning and context layer. 200K window. Holds the whole vault during synthesis. Writes drafts that sound like you.
- Obsidian - the memory layer. Plain markdown. Local. Yours. Survives every model swap.
- Kimi K2.6 / open models - the orchestration layer. Long-horizon tasks Claude doesn't have to babysit. Open-source. Free. Runs 4,000 steps without flinching.
- Cursor - the execution layer. For when output is code or product. Cloud handoff means your laptop can close.
- N8N - the automation layer. $5/mo droplet. Schedules every workflow. Routes captures. Runs the whole loop while you sleep.
The mistake almost everyone makes is using one tool for all five layers. ChatGPT for everything. Or Claude for everything.
The $30K vault uses the right tool at each layer. Claude is the brain. Obsidian is the memory. Open models scale the cheap stuff. Cursor ships when ships is required. N8N glues it.
## The Output Generator Prompt That Earns the Most
Of the 5 workflows, the Output Generator is the one that produces revenue directly. The other 4 feed it.
Here is the production version of the prompt - tuned across 6 months. Drop it into 07 - SYSTEM/prompts/output-generator.md:
The expected_revenue stamp at the bottom matters. After a few weeks, you have data on which note types and which topics produce the highest-value output. The vault starts telling you what to capture more of.
## The 6-Month Revenue Curve
This is what compounds for $30K - and it does NOT happen on Day 7.
The 6-month curve is the moat. Anyone can install Obsidian today. Almost nobody will still be capturing in 90 days.
The people who are will own a context layer no late mover can replicate.
## The Step-By-Step Build (4 Evenings, $0)
Same 4-evening build from the last article - adapted for the income engine. Total: ~6 hours.
## The Mistake That Keeps This at $0
Most people who build the vault never reach $30K because of one mistake:
They never ship from it.
They capture. They run the briefings. They tweak CLAUDE.md. They post screenshots of their graph view on X.
They never open 03 - OUTPUT/. They never run the Output Generator. They never publish.
The vault is not magic. It does not produce revenue by existing.
The vault produces revenue when you run Workflow 5 and ship the result. Once a day. Or twice a week. Or once a Sunday.
The shipping cadence is what closes the gap between "I have a vault" and "I make $35K/month."
## The Quiet Truth Underneath All Three Articles
None of them show you the shipping loop because that is the one part nobody can hand you. The structure is replicable. The capture is replicable. The tools are replicable.
The shipping discipline is not.
That is what the $35K/month operator has that the 4,000-note hobbyist does not. Not better prompts. Not a better model. Not a fancier folder structure.
A standing order with themselves to ship from the vault, every week, no matter how rough the draft.
The vault rewards that discipline disproportionately.
Six months of capturing + 6 months of shipping = a context layer AND a body of public work nobody else can copy.
That is the $35K/month moat.
## 相关链接
- [ZEUS](https://x.com/zeuuss_01)
- [@zeuuss_01](https://x.com/zeuuss_01)
- [16K](https://x.com/zeuuss_01/status/2060364984052285459/analytics)
- [$13-25K](https://x.com/search?q=%2413-25K&src=cashtag_click)
- [Upgrade to Premium](https://x.com/i/premium_sign_up)
- [10:17 PM · May 29, 2026](https://x.com/zeuuss_01/status/2060364984052285459)
- [16.8K Views](https://x.com/zeuuss_01/status/2060364984052285459/analytics)
- [View quotes](https://x.com/zeuuss_01/status/2060364984052285459/quotes)
---
*导出时间: 2026/5/30 11:05:22*
---
## 中文翻译
# 如何将 Obsidian + Claude 武器化，在 4 个晚上取代月薪 3.5 万美元的内容团队？
**作者**: ZEUS
**日期**: 2026-05-29T14:17:46.000Z
**来源**: [https://x.com/zeuuss_01/status/2060364984052285459](https://x.com/zeuuss_01/status/2060364984052285459)
## 捕获/应用差距（为什么大多数知识库一文不值）
你读了一篇有趣的文章。你高亮了一段。你把它归档到 /articles 目录下。六个月后，你搜索某个模糊相关的东西，偶然发现了这条笔记。你读了它，心想“我本应该记得这个”，然后继续前进。
那条笔记从未变成任何东西。
它从未变成一条推文。一份通讯。一个客户策略。一个定价论点。一个融资演讲稿。一篇博客文章。一个产品功能。一个谈话要点。一条咨询建议。
它就躺在那里。
这是一个结构性错误。大多数 Obsidian 设置都优化了收件箱。但出口无人值守。
- 你有 4,000 条笔记和零篇发布的文章。
- 你有 12 个捕获面和零个工作流。
- 你有一个漂亮的图谱视图，没有客户交付物。
能赚钱的知识库恰恰相反。它是从输出反向优化的。
## 四区知识库（从输出反向构建）
忘掉上一篇文章里的 9 个文件夹吧。专门为了变现，你可以把知识库压缩成 4 个区域：
不对称是关键。你一天中的大部分时间都花在 02 和 03。捕获位于 00。上下文做支持。但工作——那些可计费、可发布、可交付的工作——位于 02 项目和 03 输出。
你应该不假思索地知道本周在 03 - 输出/中生成了多少条笔记。
如果答案是零，那这个知识库就是个爱好。不是一门生意。
## 3 个捕获面
你不能从空空如也的知识库里发货。捕获层必须足够激进，让你根本不可能丢失一个想法。
三个面就能覆盖所有情况：
- 面 1 - QuickAdd（桌面端）：一个键盘快捷键。一个浮动输入框。笔记直接落入今天的日记。零阻力。捕获时不分类。Claude 稍后负责分类。
- 面 2 - 手机小组件（手握手机）：点击主屏幕小组件。打字或语音转文字。笔记在 30 秒内同步到知识库。
- 面 3 - Telegram 机器人（未拿手机）：散步时，在车里，谈话中。向 Telegram 机器人发送语音消息。落入今日日记的捕获区。
Telegram 机器人是整个技术栈中 ROI 最高的一环，因为它捕获了那些你本来会永远丢失的想法——那些在散步、洗澡、坐 Uber 时冒出来的想法。
这些想法往往就是那些能变现的。
## 5 工作流输出引擎
当 5 个工作流在上面运行时，知识库就开始产生收入。每一个都是存储在 07 - 系统/提示词/ 中的一个提示词。每一个都被调度或触发。它们共同将原始捕获转化为可交付的输出。
**工作流 1 - 夜间处理器（凌晨 2 点）**
按笔记类型对当天的捕获进行分类。将它们从 00 - 捕获/ 移动到 01 - 上下文/。添加前置数据。纯机械工作。
**工作流 2 - 晨间简报（早上 6 点）**
阅读过去 7 天的上下文。将一份 400 字的简报写入 /简报/。展示模式、矛盾、一个非显而易见的联系。
**工作流 3 - 连接面（周日）**
回顾 30 天的上下文。返回 3 个你通过主动搜索无法找到的联系。这是原创想法诞生的地方。
**工作流 4 - 项目更新器（手动，每周）**
对于 02 - 项目/ 中的每个活跃项目，阅读与该项目相关的最新捕获。更新项目文档。建议后续步骤。
**工作流 5 - 输出生成器（手动，按需）**
赚钱的工作流。你告诉 Claude 你需要交付什么。它读取 01 和 02 中的相关笔记。它用你的口吻生成一份完整的输出草稿。保存到 03 - 输出/。
输出生成器的提示词示例：
这就是那个将 4,000 条休眠笔记转化为可交付、可计费工作的提示词。
## 3 万美元到底从哪来
这里是数学明细。这些都不需要你出类拔萃——它们要求你从一个能复利增长的知识库持续发货。
两个重要的框架：
- 这些收入线都不需要某种你没有的特定技能。每一条都需要从一个能复利增长的知识库发货。
- 你不需要全部 4 项。只要稳定做到其中 2 项，凭借笔记本电脑里的一个文件夹，你就能达到 1.3-2.5 万美元/月的收入。
## 围绕 Claude 的 5 工具栈
能从知识库每月发货 3 万美元的栈有 5 个清晰的层级。每个工具都拥有其对应的一层，这是其他工具无法复制的。
- **Claude** - 推理和上下文层。20 万 token 的窗口。在综合时持有整个知识库。写出听起来像你的草稿。
- **Obsidian** - 记忆层。纯 Markdown。本地。属于你。在每次模型更迭中幸存。
- **Kimi K2.6 / 开源模型** - 编排层。那些 Claude 不需要盯着的长周期任务。开源。免费。运行 4,000 步而不退缩。
- **Cursor** - 执行层。当输出是代码或产品时。云端移交意味着你的笔记本可以合上。
- **N8N** - 自动化层。每月 5 美元的 VPS。调度所有工作流。路由捕获。在你睡觉时运行整个循环。
几乎每个人都会犯的错误是用一个工具做所有五层。所有东西都用 ChatGPT。或者所有东西都用 Claude。
3 万美元的知识库在每一层都使用正确的工具。Claude 是大脑。Obsidian 是记忆。开源模型处理廉价任务。Cursor 在需要发货时发货。N8N 将它们粘合在一起。
## 赚钱最多的输出生成器提示词
在 5 个工作流中，输出生成器是直接产生收入的一个。其他 4 个都在为它提供养料。
这是提示词的生产版本——经过了 6 个月的调优。把它放入 07 - 系统/提示词/输出生成器.md：
底部的 expected_revenue 标记很重要。几周后，你就有了哪些笔记类型和主题能产生最高价值输出的数据。知识库开始告诉你应该多捕获什么。
## 6 个月收入曲线
这就是复利达到 3 万美元的过程——而且这不会发生在第 7 天。
6 个月的曲线是护城河。任何人今天都可以安装 Obsidian。几乎没人能在 90 天后还在坚持捕获。
那些坚持下来的人将拥有一个后来者无法复制的上下文层。
## 逐步构建（4 个晚上，0 成本）
与上一篇文章相同的 4 个晚上构建——但已适配收入引擎。总计：约 6 小时。
## 让这一切维持在 0 美元的那个错误
大多数构建了知识库的人从未达到 3 万美元的收入，因为一个错误：
他们从未从中发货。
他们捕获。他们运行简报。他们调整 CLAUDE.md。他们在 X 上发布图谱视图的截图。
他们从未打开 03 - 输出/。他们从未运行输出生成器。他们从未发布。
知识库不是魔法。它不会因存在而产生收入。
当你运行工作流 5 并发布结果时，知识库才会产生收入。一天一次。或者一周两次。或者一个周日一次。
发货的频率是关闭“我有一个知识库”和“我一个月赚 3.5 万美元”之间差距的关键。
## 所有三篇文章背后的潜藏真相
它们都没有向你展示发货循环，因为那是没人能交给你的部分。结构是可复制的。捕获是可复制的。工具是可复制的。
发货的纪律不可复制。
这就是月入 3.5 万美元的从业者拥有而拥有 4,000 条笔记的爱好者所没有的东西。不是更好的提示词。不是更好的模型。不是更花哨的文件夹结构。
而是一条他们给自己的铁律，即每周必须从知识库发货，无论草稿多么粗糙。
知识库会对这种纪律给予不成比例的回报。
6 个月的捕获 + 6 个月的发货 = 一个别人无法复制的上下文层和一整套公开作品。
这就是月入 3.5 万美元的护城河。
## 相关链接
- [ZEUS](https://x.com/zeuuss_01)
- [@zeuuss_01](https://x.com/zeuuss_01)
- [16K](https://x.com/zeuuss_01/status/2060364984052285459/analytics)
- [$13-25K](https://x.com/search?q=%2413-25K&src=cashtag_click)
- [升级到 Premium](https://x.com/i/premium_sign_up)
- [10:17 PM · May 29, 2026](https://x.com/zeuuss_01/status/2060364984052285459)
- [16.8K Views](https://x.com/zeuuss_01/status/2060364984052285459/analytics)
- [View quotes](https://x.com/zeuuss_01/status/2060364984052285459/quotes)
---
*导出时间: 2026/5/30 11:05:22*