Creator OS
37studioai video automationclaude codecontent workflowopen source

AI Video Posting Automation: 37studio Posts to 9 Platforms

37studio is an open-source AI video posting automation tool where Claude Code makes on-brand videos, posts to 9 platforms, and drafts replies. Here is how to self-ho

Creator OS · October 26, 2026 · 8 min read

AI video posting automation means an AI agent renders a clip, publishes it, and drafts replies. 37studio is an open-source, self-hosted project built around Claude Code that covers that loop. Creator OS is the hosted side: a web app, REST API, CLI and hosted MCP server for 11 platforms plus Skool, so publishing runs through a connector instead of infrastructure you maintain.

  raw clip
     |
     v
 [ Claude Code ]  -- editing skills render the file
     |
     v
  finished video (med_ id)
     |
     v
 Creator OS MCP / CLI  -- one post, many networks
     |
     +-- Instagram  TikTok  YouTube  X  Facebook
     +-- LinkedIn   Threads  Reddit   Slack
Render on your machine, publish through one connector.

What Is 37studio and How Does AI Video Posting Automation Work?

37studio is a self-hosted project where Claude Code turns a raw clip into an on-brand video, posts it to nine platforms and drafts replies, using keys you supply.

The loop has three parts. First, an editing step: the agent cuts, captions and styles the footage. Second, a publishing step: the finished file goes out to every connected network. Third, an inbox step: the agent reads new comments and writes draft answers for you to approve.

The source repo is Shaitrish/37studio. Because it is self-hosted, you run the job queue, store the tokens and pay the model bill yourself. Nothing leaves your machine except the calls to each network’s API.

If you would rather not run infrastructure, split the two halves. Keep the editing on your own machine with OPEN VIDEO EDIT, then publish and reply through Creator OS, where the hosted MCP server gives an agent about 160 tools for posts, analytics, comments, DMs and automations. The publish half becomes a connector; the render half stays local.

Get started for $0

How to Get Claude Code and Set Up the Editing Side

Install Claude Code first, then decide what runs on your machine and what runs on a hosted service.

  1. Install Claude Code from Anthropic’s own documentation and sign in. The command line tool is what runs the editing skills.
  2. Clone the 37studio repo and read its setup notes for the environment variables it expects.
  3. Follow the repo’s own configuration instructions for whatever credentials it asks for, and keep that file out of version control. A self-hosted setup means the tokens are yours to protect.
  4. Run one small test: a single clip, rendered locally, watched by hand, before you batch anything.
  5. Only then connect a publishing path.

On Creator OS the account step is a sign-in, not a credential file. Connect platforms once through the authentication docs, then create a key under Settings > API keys and hand it to the agent:

claude mcp add --transport http creatoros https://mcp.creatoros.ca/mcp \
  --header "Authorization: Bearer cos_live_..."

That key is pinned to one workspace, so an agent holding it cannot publish to another client’s accounts. You can also issue a read-only key, and the API refuses writes from it.

How 37studio Makes On-Brand Videos Without Opus Clip

Instead of a template library, 37studio hands the styling job to the agent: you describe the brand, it applies the same rules to every clip.

A worked example. Write a short brand file with your fonts, caption position, colour codes, logo placement and the phrases you never want spoken. Then prompt the agent in plain language:

Edit ./raw/ep12.mp4 as a 45 second vertical short.
Use the branding in brand.md. Caption every sentence.
Keep the hook in the first 2 seconds. Export 1080x1920.

The output is a finished file, not a timeline you have to scrub. Run the same prompt on next week’s clip and the look stays consistent because the brand file did not change.

Creator OS has an open-source editing counterpart called OPEN VIDEO EDIT: Claude Code skills that turn a talking-head clip into an animated short with captions, logos, faces, GIFs, b-roll cards and counters, rendered locally with Python and ffmpeg. Skills include split-animated-talking-head, fireship-style-short and reverse-engineer. See the OPEN EDITS update and the launch video.

Claude Code Video Editor GitHub Setup: Repo Walkthrough

Treat the repo as three folders to inspect before you run anything: the skills, the publish layer and the config.

  • Skills. Markdown instruction files the agent loads on demand. Read them; they are the actual behaviour, and you can edit them for your brand.
  • Publish layer. The code that calls each network’s upload endpoint. Check which platforms are wired and what metadata each one requires.
  • Config. Where your keys, default caption style and brand rules live. This is the file you back up.

Two practical checks before you trust it with a real account. Post to a test account on the network with the fussiest rules first, because a rejection there usually means missing metadata rather than a broken pipeline. And log every response body, since each network returns a different error shape.

Creator OS standardises that last point: every error comes back as { error: { code, message, status } }, and ids are typed tokens like post_, cmt_ and med_ so you always know what you are passing. The posts API docs cover the request shapes.

How to Auto Post One Video to 9 Platforms From Claude Code

Upload the file once, get a media id back, then publish to every network in a single call so each platform’s rules are applied for you.

creatoros media:upload ./exports/ep12-short.mp4

creatoros posts:create \
  --text "The 3 edits that fixed my retention" \
  --platforms instagram,tiktok,youtube,x,facebook,linkedin,threads,reddit,slack \
  --media med_9f2c \
  --scheduledAt 2026-03-04T17:00:00Z --timezone America/Toronto

Three things make this safer than nine separate scripts. The media id is reusable, so the file uploads once. Length is validated before anything is sent: creatoros validate:post-length --text "..." flags captions that break a platform’s limit. And if a network is named but not connected, the response lists it in missing_platforms while the rest still go out.

For per platform settings, use the advanced form of POST /v1/posts, where each platform is an object carrying its own options. A TikTok draft, for example, goes to the creator inbox to finish in the app:

{
  "content": "Hook, story, payoff",
  "platforms": [
    { "platform": "tiktok", "content": "Hook, story, payoff" }
  ],
  "media": ["med_9f2c"],
  "tiktok": { "draft": true }
}

Ask the agent for the round trip and it will report back: list_posts to confirm what went out, get_post_analytics to pull numbers, get_best_time_to_post to pick the next slot. There is a full walkthrough in auto posting one video to 11 platforms, and more on the KevBuildsApps YouTube channel.

Drafting Replies and Comments: Agent-Driven Community Management

Publishing is half the job: the agent should also read new comments and write draft answers you approve, never post blind.

In 37studio this is a drafting loop inside the same session. On Creator OS it is a tool call with a hard boundary. list_comments returns the new ones, the agent writes a reply, and you approve before reply_to_comment or hide_comment runs. Destructive tools are marked so the AI app asks first.

Task CLI MCP tool
Pull new comments inbox:comments --platform instagram list_comments
Reply publicly inbox:reply <postId> --accountId <acc_> --message "..." reply_to_comment
Send a DM after a comment inbox:private-reply <postId> <cmt_> --accountId <acc_> --message "..." create_automation
Check DMs inbox:conversations --status active list_conversations

Keyword funnels are where this pays off. A create_automation call matches a word in a comment and sends the resource by DM, which turns replies into leads without you watching the inbox. Worked example: keywords template,swipe, a DM with the link, and a short public reply so other readers see the answer exists. The comments to DM guide covers the setup.

37studio vs Buffer vs Opus Clip: Self-Hosted Tradeoffs in 2026

The real choice is who runs the machine: you, or a vendor.

A social media scheduling dashboard like Buffer puts publishing behind a web interface you subscribe to. A clip editing tool like Opus Clip processes video on its own servers. 37studio is the other shape: the code sits in your repo, the agent runs locally and your keys never leave your infrastructure.

37studio Creator OS
Where it runs Your machine or server Hosted web app, iOS app, hosted MCP server
Credentials As configured in the repo Connect accounts once, then a scoped cos_live_ key
Agent access Claude Code skills in the repo About 160 MCP tools, REST API and CLI
Networks Nine, as wired in the repo 11 platforms plus Skool, WordPress and Shopify
Video rendering Yes Not through the MCP server; use OPEN VIDEO EDIT
Maintenance Yours Vendor’s

Self-hosting wins when you want to read and change the code. Hosted wins when you want to spend the day making videos, not patching upload handlers. Creator OS ships a hosted MCP server and a CLI so an AI agent can run your accounts, and a Claude Code plugin adds a publish guard that asks Proceed or Cancel before anything is published, spent or deleted.

FAQ: AI Video Posting Automation With Claude Code

How to get Claude Code?

Install it from Anthropic’s own documentation and sign in with your Claude account. Then add Creator OS as a connector with the claude mcp add command in this post, or browse the Claude Code plugin docs.

Is 37studio a free alternative to Buffer and Opus Clip?

37studio is open source, so there is no licence fee, but you pay for model usage and hosting, and you configure it with your own accounts. On Creator OS, plans start at $14.99/month for 11 platforms plus Skool, with the API, the MCP server and the CLI included.

Claude Code video editor GitHub repo setup?

Clone Shaitrish/37studio, read the skills folder to see what the agent actually does, and follow the repo’s configuration notes for the accounts it publishes to. Test one post to one network before running the nine platform job.

Can Claude Code post videos to 9 platforms automatically?

Yes, through either path. 37studio handles it in the repo, and Creator OS handles it with one posts:create call or the create_post MCP tool, which applies each platform’s rules and returns any unconnected network in missing_platforms.

Does 37studio draft replies to comments automatically?

Yes, it drafts replies as part of the same workflow. On Creator OS you can go further with keyword automations that match a word in a comment and send a DM, on Instagram and Facebook.

Get started

You do not have to choose between a repo you maintain and a dashboard you cannot touch. Start on the hosted side: sign up, connect your accounts, create an API key and add the MCP server to Claude Code. Creator OS plans start at $14.99/month for 11 platforms plus Skool, with the API, the MCP server and the CLI included. Read the MCP docs and the posts API docs first if you want to wire it up yourself.

Get started for $0
Free trial for new subscribers. Cancel anytime.

Keep reading