# How To Build Own Content Engine? (FULL COURSE)
**作者**: Ronin
**日期**: 2026-04-10T14:03:04.000Z
**来源**: [https://x.com/DeRonin_/status/2042604279077237170](https://x.com/DeRonin_/status/2042604279077237170)
---
I run 10 social medias for my clients and don't write a single post manually
No content team. No $8-12k/mo agency retainer. No sitting in front of ChatGPT rewriting the same damn post ten times for ten platforms
Just a folder of .md files + one AI agent, and a system that takes one idea and produces 10 platform-native posts (each one actually thinking about the topic differently)
This article is the full breakdown I promised. Everything what I use in my daily workflow is here. Every file, every wikilink, every node
By the end you'll have a complete content production system you can realistically build this WEEKEND
After reading and implementing this, you'll have:
FULL Automated Content Engine
An AI agent that actually understands your voice, audience, and platform rules
A repurposing engine that turns one topic into 10 ready-to-publish posts
A working skill graph folder with every file filled in and connected
5. A content system that runs 10 accounts while you go live your life (across 10 different social platforms)
[ Let's build it ] ↓↓↓
## What Is a Skill Graph? (And Why a Folder of .md Files Replaces an Entire Content Team)
Most people use AI for content like this:
1: Open Claude
2: Type "write me a LinkedIn post about productivity"
3: Get a generic post that sounds like a corporate intern wrote it
4: Spend 20 minutes making it not sound like a robot
5: Then do the whole thing again for every platform
That's not a system. That's a chore with extra steps
The problem isn't the AI. It's that you're giving it ZERO context about your brand, your audience, your voice, your platform strategies, or how any of this connects
You're basically hiring a genius with amnesia every single time you start a new chat
SOLUTION: a skill graph fixes this
Think of it like hiring a new employee
You could throw them into the deep end on day one with no onboarding, no documentation, no context about how your company works and just hope for the best
Or you could hand them a complete playbook that explains who you are, how you operate, who you're talking to, and what good output looks like
A skill graph is that playbook. Except it's for your AI agent
Technically it's just a folder of interconnected markdown files where is each file is one "knowledge node", one piece of your content system's brain
Inside each file you use [[wikilinks]] (double-bracket links like [[brand-voice]] or [[hooks]]) to reference other nodes
When you point an AI agent at this folder and give it a topic, it doesn't just read one file. It follows the links, reads the connected nodes, and builds up a complete understanding of your brand, voice, audience, platform rules, hook formulas, and repurposing logic BEFORE writing a single word
The difference is insane:
BAD: A single prompt = hiring a freelancer with no brief, no brand guidelines, no knowledge of your audience
GOOD: A graph of 30+ interconnected .md files = hiring a full content team that's read your entire playbook and knows exactly how each platform works
Another way to think about it: one flat .md file gives you a TOOL (a simple reference doc)
A graph gives you a TEAM (a system with sub-specialists for every platform, hook type, voice variant, and audience segment)
This is what replaced $5K/month in content production costs for me
Not because the individual files are magical, but because the connections between them create intelligence no single prompt can match
[ PLAYBOOK BEGINS ] ↓↓↓
## Where to Build It? (Your Tools Stack)
You have two options. Both work fine
1. Obsidian — this is what I'd recommend if you want to SEE the graph visually
Obsidian is a free markdown editor that natively supports [[wikilinks]] and has a gorgeous graph view that shows you how all your nodes connect
It's like looking at a neural network of your own content system. Makes debugging super intuitive too, you can see which nodes are disconnected and which ones are overlinked
2. A regular folder on your desktop — if you don't want another tool (tool fatigue is real), a plain folder of .md files works perfectly fine
The AI agent doesn't care about Obsidian, it reads the markdown and follows the [[wikilinks]] regardless of what editor you used. VS Code, Notion exported to markdown, even Notepad. Literally anything works
Which tools actually run the system:
- Claude is my primary choice (the best brain). Claude Projects lets you upload all your .md files and keep them as persistent context, so every conversation in that project has access to your full graph
- ChatGPT works with custom GPTs or by pasting key files into the conversation
- Cursor is if you're technical and want the agent to read files directly from your local file system
Pick anything you're comfortable, but I guess 95% of readers will definitely pick Claude (and that's the best choice imo, I use it)
## THE FOLDER STRUCTURE
Here's the exact structure you're gonna build:
17 files. 4 folders. That's your entire content production machine
Go create this structure RIGHT NOW. Seriously, pause reading and do it!!!
Step-by-step:
- Open Obsidian or just make a folder on your desktop
- Create the subfolders
- Make empty .md files with the names above
- Then come back and we'll fill every single one
Quick overview of what each folder does before we go file by file:
index.md — the entry point. The briefing document your AI agent reads first. Most important file in the entire system
platforms/ — one file per platform with the rules, format, character limits, posting frequency, content style. Everything the agent needs to write NATIVELY for that platform
voice/ — your brand voice DNA and how it adapts per platform. This is what stops your content from sounding like a robot generated it
engine/ — the operational backbone. Hook formulas, the repurposing chain, scheduling rules, content type definitions
audience/ — who you're actually talking to. Different audience segments get different angles on the same topic
[ Now let's fill every file ] ↓↓↓
## File 1: index.md (The Command Center)
This is the single most important file in your entire graph
Every time you give your AI agent a topic, it starts here. If this file is weak, everything downstream will be weak. Think of it like the CEO of your content operation, it sees the full picture and delegates to specialists
index.md is NOT a table of contents. This is the most common mistake I see people make. It's not a file list
It's a BRIEFING. It tells the agent who you are, what the system does, and exactly how to execute
Put three things in it:
A few things to notice about this file:
The identity section isn't filler. The AI actually uses it to calibrate everything it produces
"AI automation, SaaS building, and monetizing tech skills" gives you completely different content than "vegan meal prep for busy parents". Be SPECIFIC about your niche here, this part is THE MOST IMPORTANT!!!
The node map gives context with every link. Not just "[[x]] — Twitter" but "[[x]] — short-form, hook-driven, 280 chars max, casual lowercase. post 5x/week minimum"
That extra context helps the agent make decisions without needing to open every single file for every task. Saves tokens too
To summarize, this file moderates the whole process of content production in this system
Please, be careful with it and put the biggest focus here
## File 2: x.md (X Playbook File)
Each platform file is a complete playbook for producing content on that specific platform
I'll show X as the detailed example (every other platform file follows the same structure, you just swap the specifics):
## File 3: linkedin.md
## File 4: instagram.md
## File 5: tiktok.md
[HOOK: 2 seconds (text overlay + voiceover)]
[CONTEXT: 5 seconds]
[THE HOW: 30-40 seconds]
[RESULT: 5 seconds]
[CTA: 3 seconds]
## File 6: youtube.md
[HOOK — 0:00-0:30]
State the problem + promise the solution. "I run 10 social media accounts and don't write a single post manually. By the end of this video you'll know how to build the exact same system."
[CONTEXT — 0:30-2:00]
Why this matters. What most people get wrong. Set up the value
[MAIN CONTENT — 2:00-9:00]
Step-by-step walkthrough. Screen recordings, demos, examples. Break into 3-5 clear sections with timestamps
[RECAP + CTA — 9:00-10:00]
Summarize key points. CTA: subscribe or grab the template
## Files 7-9: threads.md, facebook.md, newsletter.md
These follow the same structure as above. I'll give condensed versions since the pattern should be obvious by now:
[ Setup Your DNA Below ] ↓↓↓
## File 10: brand-voice.md (Your DNA)
This file defines who you are across ALL platforms
It's the source of truth every other file references. Think of it like your content DNA (the platforms adapt the expression, but the underlying identity stays the same):
Also, one of the most important parts and probably you will have to apply some changes as it will start generating the drafts
But don't worry if you build this system around Claude, since it will rework all changes automatically inside of your project
## File 11: platform-tone.md (Adapt DNA To Each Platform)
This is one of my favorite files in the whole system. It's the bridge between your universal voice and each platform's culture
Like how you talk differently at a house party vs a business dinner vs a podcast interview (same person, different energy):
## File 12: hooks.md (Create Hooks Patterns)
Hooks are where 80% of your content's performance is decided
You could write the most incredible post ever, but if the first line doesn't stop someone mid-scroll, NOBODY will ever read it
It's like having a perfect restaurant in a back alley with no sign
CTRL C + CTRL V:
The right hooks define 70% success of your tweet/post/video
Pay a lot of attention here's too
## File 13: repurpose.md (1 Idea = 10 Different Platforms)
OK this is the file that makes the whole thing actually work
Without this, you just have a bunch of reference docs. WITH this, you have a production pipeline
One idea goes in, ten platform-native posts come out:
## File 14: scheduling.md (Content Calendar)
## File 15: content-types.md (Separate Formats List)
## Files 16-17: builders.md and casual.md (Identifier Of Your Audience)
These 2 files identify to which audience this type of content will fit the best and if you want to touch some exact one, it just recreate an idea and adapt exactly under this audience
## How to Actually Use This Thing???
Alright, you have all the files. Let me show you how to plug it in and start producing:
Method 1: Claude Projects (what I'd recommend)
Create a new Project in Claude called "Content Skill Graph"
Upload all 17 files into the project knowledge base
Start a new conversation and give it a topic:
4. Claude reads index.md, follows the wikilinks, reads platform files, voice files, hooks, repurposing chain, and outputs 8 platform-native posts
5. Review. Schedule. Publish. Go do something else with your day
Method 2: Paste Context (simplest, works with anything)
If you don't have Claude Pro or prefer another tool:
Copy the contents of index.md
Paste it into any AI chat (Claude free, ChatGPT, whatever)
Add: "Here's my content system. When I give you a topic follow the execution instructions. For each platform write a native post that rethinks the topic for that platform, not reformatted, completely rethought"
Give it topics
For better results also paste brand-voice.md and whichever platform files you're producing for. More context = better output. Always
Method 3: Cursor / Claude Code (most powerful, most technical)
Keep the skill graph folder on your local machine
Point Cursor or Claude Code at the folder
The agent reads files directly from your file system
It can also UPDATE the files by adding new hooks to hooks.md, refining platform-tone.md based on what's performing
This is the fully autonomous version where the graph literally evolves itself over time. It's like giving the system a memory that compounds
## What the Output Actually Looks Like (This Is Where Most People Mess Up)
Let me show you what "rethinking, not reformatting" actually means in practice. Because this is the part that separates a skill graph from a copy-paste machine
Say your topic is: "How I use AI to manage 10 social media accounts"
Here's what each platform version looks like:
X: contrarian thread, lowercase casual, step-by-step. "you don't need a content team. you need 30 markdown files. here's how I run 10 accounts without writing a single post manually:"
LinkedIn: personal narrative, professional tone, ~1,500 words. "6 months ago, I was spending $8,000 a month on content production across 10 platforms. Today I spend $0. Here's exactly what changed."
Instagram: 7-slide carousel, visual-first, bold claim on slide 1. Slide 1: "I Run 10 Accounts And Don't Write Anything." Slides 2-7: the system broken into visual steps. Slide 8: "Save this. Follow for more"
TikTok: 45-second raw screen recording script. "You're still writing content manually for every single platform? Let me show you what I use instead." [shows the folder structure, Obsidian graph view, Claude spitting out posts]
YouTube: SEO title + structured outline, 8-minute format. "How to Run 10 Social Media Accounts with AI (Complete System Walkthrough)" — full tutorial with screen recording
Newsletter: 1,500-word deep dive. "This week I want to pull back the curtain on something I've been building quietly for months..."
Threads: hot take, conversational. "hot take: the future of content isn't AI that writes for you. it's AI that thinks like 10 different people for you"
Facebook: community discussion. "Has anyone else tried building a system to manage multiple social accounts at once? Here's what I've been experimenting with — curious what you all think"
Same topic. Eight completely different pieces of content. Each one native to its platform. Each one valuable even to someone who follows you literally everywhere
Just provide the topic and get 8 different angles at once
## CONCLUSION (what to do)
Create the folder. 17 files, 4 folders. Takes 2 minutes. Seriously, do it right now
2. Fill in index.md and brand-voice.md first. These two define everything else. If you don't know who you are and what your system does, the other files won't matter
3. Fill in the platform files for your top 3 platforms. You don't need all 8 on day one. Start with the 3 where you're most active. Add the rest later
4. Fill in hooks.md and repurpose.md. These are your engine. How you start posts and how you multiply them
5. Upload everything to Claude Projects (or paste into whatever AI tool you use)
6. Give it a topic and test. See what comes out. Adjust the files based on the output. First version won't be perfect and that's fine, it gets better over time you refine the nodes
7. Iterate weekly. Update hooks.md with what's performing. Refine platform-tone.md as you learn what sounds right. Add new platform files when you're ready
The skill graph is designed to grow. Start small, add nodes, refine connections. Every week it gets smarter because you're encoding what you've learned into the files themselves
It's literally like compound interest but for your content system
One folder. 17 markdown files. One AI agent. 10 platforms. ZERO manual writing
That's how my current content engine looks like and I guess in few weeks I will make it even more adaptive and will improve research part
If you loved this article and you found it interesting, please, put the LIKE + RT ❤️
If this article collects 2,000+ Likes, I release a detailed video guide + my "Research Engine" in one more article
Now, it's time to apply to your system buddies.
## 相关链接
- [Ronin](https://x.com/DeRonin_)
- [@DeRonin_](https://x.com/DeRonin_)
- [824K](https://x.com/DeRonin_/status/2042604279077237170/analytics)
- [Upgrade to Premium](https://x.com/i/premium_sign_up)
- [10:03 PM · Apr 10, 2026](https://x.com/DeRonin_/status/2042604279077237170)
- [824K Views](https://x.com/DeRonin_/status/2042604279077237170/analytics)
- [View quotes](https://x.com/DeRonin_/status/2042604279077237170/quotes)
---
*导出时间: 2026/4/15 00:33:17*