Skip to content

Guides

From installto your first post.

Short guides. Copy, paste, done.

About 20 minutes to set up

01

The easy way: let AI set it up

Open the folder, start your AI, paste one prompt.

  1. Accept the GitHub invite, then clone the repository it shows you.
  2. Open the contentos folder: in VS Code, File → Open Folder.
  3. Start your AI there: the Claude Code panel in VS Code, or claude, codex or gemini in its terminal.
  4. Paste the prompt below and answer its questions.
cd contentosclaude

Setup prompt· Paste into Claude Code, Codex or Gemini

Set up Content OS on this computer. The project is this folder. 1. Check Node.js 22+, pnpm, Git and FFmpeg. If one is missing, tell me the command and install it after I say yes. 2. Run pnpm install, then pnpm exec playwright install chromium. 3. Copy .env.example to .env.local. Leave the keys empty; I will add mine. 4. Run pnpm cos doctor and fix every ✗ it shows. 5. Start the Studio with pnpm dev and give me the link. One short line per step. Stop and ask before anything that needs my password or costs money.

Content OS ships with CLAUDE.md, AGENTS.md and GEMINI.md, so Claude Code, Codex and Gemini all follow the same rules.

02

Quick start

Rather do it yourself? Six lines.

cd contentospnpm installpnpm exec playwright install chromiumcp .env.example .env.localpnpm cos doctorpnpm dev

Then open localhost:3000/studio, and open the folder in VS Code with Claude Code.

The next four guides walk through each line.

03

Get the code

A private GitHub repository, yours for life.

  1. Buy Content OS.
  2. In your account, connect your GitHub.
  3. Accept the invite GitHub emails you.
  4. Clone the repository it opens, then open the folder.

04

Install the tools

Once per computer. Open a new terminal after.

ToolHow to get it
Node.js 22+nodejs.org, or brew install node
pnpmcorepack enable
Gitgit-scm.com, or xcode-select --install
FFmpegbrew install ffmpeg, or winget install Gyan.FFmpeg
Claude CodeVS Code → Extensions → “Claude Code”

Optional: yt-dlp for video research, Whisper for local transcripts.

05

Add API keys

All optional. No key, that power stays off.

cp .env.example .env.local
KeySwitches on
OPENAI_API_KEYAI images
ELEVENLABS_API_KEYVoices, music, sound effects, transcripts
FISH_AUDIO_API_KEYYour cloned voice
SNAPGEN_API_KEYAI video
FAL_KEYPhoto to 3D model

Or add them later in Studio → Settings. Keys never go into git.

06

Check and start

One command tells you what’s missing.

pnpm cos doctor

Fix every ✗ it shows. Then start the Studio:

pnpm dev

Studio: localhost:3000/studio. Claude Code: open the folder in VS Code. It reads the rules on its own.

07

Your first page

One page per brand. Pages never mix.

/new-page "My Brand" en instagram
  1. Answer its questions: who you are, who you talk to, how you sound.
  2. Add your logo, colours and fonts with /brand.
  3. Fix any one part later with /context, like /context my-brand voice: …
  4. Open Studio → Context to see exactly what the AI reads.

First page prompt· Paste into Claude Code, Codex or Gemini

Make a new Content OS page for my brand. Name: <your brand> Language: English Main platform: Instagram Logo and photos: <folder with your files> Ask me one question at a time about my audience, my voice, what I sell and my rules. Then set up the brand kit and show me the page in the Studio.

08

A day with Content OS

Ask, check, export. That’s it.

  1. /next-post my-brand: get ideas, each with a reason.
  2. /carousels my-brand from-idea <id>: make the one you like.
  3. Check it in the Studio. Not right? /improve my-brand <id> --fix-qa
  4. /export my-brand <id>: media, caption and hashtags, ready to post.

Log real work with /experience. Teach it a rule once with /teach.

Daily prompt· Paste into Claude Code, Codex or Gemini

What should <my-brand> post today? Give me three ideas with a reason each. Then make a carousel from the one I pick, run QA, and export it.

09

Make a video

Script, voice, scenes, render. One line.

/video my-brand "launch day" 45s reel
  1. It writes the script in your voice.
  2. It picks a format and builds the scenes.
  3. It renders on your computer and checks timing and audio.
  4. Watch it in Studio → Videos, then Render final.

Want the format demos on your machine? pnpm cos demos render (about an hour).

10

Every shortcut

Plain words work too. These are just faster.

You typeYou get
/next-post <page>Ideas, each with a reason
/carousels <page> 10Ten carousels, checked and rendered
/post <page> linkedin-post "<topic>"A text post or one graphic
/video <page> "<topic>" 45s reelA finished video
/repurpose <page> <id> --to x-threadOne idea, for another platform
/improve <page> <id> --fix-qaFixes what QA found
/teach <page> "<rule>"A rule it keeps for good
/experience "<what happened>"Your real work, saved as material
/newsNews with a business angle
/research <page> "<question>"Answers with sources
/qa <page> --all --fixChecks everything, fixes what fails
/export <page> <id>A ready-to-post package

11

Work as a team

Optional. Your own server, your own rules.

  1. Host the sync server (it ships in the repo) on your own server.
  2. Connect your computer first: that makes you the owner.
  3. Invite people as editor or viewer, page by page.

Only files sync. Keys stay on each computer. Conflicts are kept, never overwritten.

12

Get updates

Every update, for life.

git pullpnpm install

13

If something breaks

The usual fixes.

ProblemFix
ffmpeg not foundInstall it, then open a new terminal
“Browser not found” on renderpnpm exec playwright install chromium
Studio shows an old pageStop pnpm dev, delete .next/, start again
Format cards show text, not videopnpm cos demos render
An ability says [OFF]Its key is missing, or it’s off in Settings

Stuck?

Run pnpm cos doctor first. Still stuck? Ask Claude Code: it knows the whole project.

Get Content OS · $119Features