Operator Cmo

Operator CMO

**An AI marketing team that has to show its plan.**

A Claude Code plugin: eight skills in three layers, built so that no asset can be produced without naming the decision it came from.

[](https://docs.claude.com/en/docs/claude-code/plugins) [](LICENSE)

---

Why this exists

In August 2026 [Ignacio Prieto](https://www.linkedin.com/posts/ignacioprietom_i-could-not-afford-a-cmo-so-i-built-one-share-7497741014326009857-7mlJ/) posted that he could not afford a CMO, so he built one inside Claude: seven modular skills, each a plain file, each owning a job that used to take a person. It went wide, and the best comment under it was a question:

"These skills are execution of a marketing plan. Who made the plan?"

That is the whole gap. Seven execution skills replace a marketing ops team, not a CMO, because the job a CMO actually does is choosing: who we are for, what we will not do, and what we stop when the number does not move. None of that is execution, and none of it was in the stack.

The second criticism was just as sharp. If a human approves every output, and the agent produces faster than the human reads, approval becomes the constraint. You have not removed the bottleneck, you have moved it into a queue.

Operator CMO is the answer to both. The plan is a file. Execution skills refuse to run without it. And the approval step is a tiered policy with a throughput cap, not a promise to read everything.

Full write-up: [An AI CMO without a plan is just a very fast intern](https://wojciech.io/insights/ai-cmo-claude-skills-marketing-system/).

The three layers

Layer	Skill	What it owns
Decide	positioning	ICP, the problem in the buyer's words, proof, offer, anti-goals
	quarter	Two or three bets, one metric each, a kill condition a