Create account

DEV Community

sayoni dutta roy

Posted on Oct 21

Open source skills that turns Claude into an AI CMO

#ai #claude #opensource #showdev

Don't feel like reading? Here's the repo:

SupercmoHQ / superCMO-skills

Open-source skills that empower any AI agent (Claude, Cursor, Hermes, etc.) to generate end-to-end marketing campaigns - UGC videos, ad videos, product photography and more.

SuperCMO Skills

Enable AI agents to create marketing videos and images

Open-source skills that empower any AI agent (Claude, Cursor, Codex, Hermes, etc.) to generate end-to-end marketing campaigns - UGC videos, ad videos, product photography and more.

Just provide a product photo and a brief. The agent extracts the product details, selects the optimal AI image and video models, casts AI actors, and edits the generated content into finished, campaign-ready assets. It can produce videos of any length while maintaining perfect actor and product consistency.

The open-source alternative to closed AI marketing agents.

Install in one command — into Claude Code, Cursor and Codex, whichever you have (needs uv):

npx --yes github:SupercmoHQ/superCMO-skills --all

For Cursor, run it in your project with --cursor --project-dir . so the skills land there too.

Then ask your agent to "set me up": bring your own model keys, or sign in and pay per…

View on GitHub

Marketing requires a whole team effort

A proper marketing process involves research, strategy, and production teams that film, edit, and prepare content for all the platforms. Most small brands don't have any of these specialists. They usually have a founder, an e-commerce website, and at best, one evening per week for marketing.

So, we built SuperCMO Skills: open-source skills that transform Claude Code, Cursor, or Codex into an AI CMO that does everything from the beginning of ideation to the launch-ready campaign.

How it works

You interact with it just like you would with a CMO.

1. "Set me up from mybrand.com."

The AI scans your website and gets to know your brand's voice, target audience, product or service offer, visual style, and competition.

2. "What ads are my competitors running, and what's working?"

A vision model analyses competitors' image and video ads in full. It finds out which competitors' ads are the longest and therefore the most successful, which were scrapped, and what angles aren't explored in the category yet.

3. "What should my next campaign be?"

It creates ranked campaign ideas based on the testability of each one. Every idea will specify who the ideal buyer is, the key hypothesis, and the proof points.

4. "Create the assets"

It creates the necessary image and video assets - UGC videos, product videos, image ads and more.

Specialized skill for the latest ad formats

SuperCMO is equipped with specialized skills for every trending ad format:

UGC videos: Produces UGC videos of any length - like review, unboxing, try-on, tutorial and more.
Product and brand videos: Produces a polished, cinematic product commercial in your brand's voice.
Cartoon/animated ads: Produces a drawn, animated video for your product — cartoon, anime, illustrated or painted.
Image ads: Produces scroll-stopping image ads for your product.
Product photography: Directs a commercial photoshoot around your product, automatically applying the correct lighting, framing, camera angles and props to deliver studio-quality product shots across 10+ formats.
Ad cloning: Recreates a video ad you admire with your own product.

Deals with the difficulty in production

Generating a single AI clip is easy; managing the surrounding pipeline is hard. SuperCMO automates the heavy lifting:

Picking the model: It routes each shot to the best AI image or video model.
Writing the prompts: It translates ideas into the specific prompt syntax each model expects.
Casting the actor: It generates a photorealistic AI actor tailored to your buyer persona.
Keeping characters consistent: It storyboards and locks reference images to ensure the same actor and product appear in every shot without face drift.
Finishing the video: It writes the script, records voiceovers, adds captions, and stitches the clips together.

Since AI videos can be costly to produce, the agent gets your consent before creating any expensive asset. With every generation, there is a dry_run setting where you receive an exact copy of the API call (keys masked).

How it's built

The system relies on two components:

The skills: Plain SKILL.md files handling specific jobs. The agent loads the right skill for your request, and the skills pass context to each other (brand → research → plan → assets → formats).
A local Python MCP server (supercmo-skills on PyPI) that executes the image, video, and voice API calls. Your agent starts it on demand with uvx, so there's nothing to pip install.

You can bring your own vendor keys (requests go straight to the vendor with zero markup) or pay per use with our keys. There are no subscriptions.

Install (requires uv):

npx --yes github:SupercmoHQ/superCMO-skills --all

In Cursor, run it inside your project with --cursor --project-dir . to land the skills in your directory. Then, simply ask your agent to "set me up."

Try it