The whole GTM playbook in 12 skills, wired to cascade, not scattered prompts

Last week I went through a run of repos that bundle business skills. Most looked alike: a box of good prompts you call one at a time. This one is strong at that, that one is strong at another.

But one made me stop and look for a while. It is called gtm-strategist-skills. It does not just pile up good prompts, it lays out up front how each piece of work will connect to the next.

The owner is Maja Voje, who wrote the book GTM Strategist. She took the methodology from the book, which she says has been tested with 1000+ companies, and released it as an open-source Claude skill set that is free to use. That is the interesting part, not just as a GTM tool, but as an example of design: how do you take one person's expertise and make it something an AI can carry on with?

I will tell this in three parts. First, what it is. Then the design that matters more than the frameworks. And finally the lessons you can use even if you are not doing GTM at all.

Part 1What gtm-strategist-skills is

It is 12 Claude skills that trace the whole go-to-market path, 100 sub-tasks in total. The author groups them into four broad stages.

• Understand, first lay the foundations (OPE canvas, SWOT, value proposition, a 90-day plan), gather market evidence, find the beachhead, interview customers, analyze competitors, then test the assumptions against real customers until you have a validated ICP (ideal customer profile).
• Build the product and set pricing a value proposition along JTBD lines (Jobs to be Done, the jobs a customer wants done), a roadmap, an MVP, and pricing set by looking at both competitors and willingness to pay (Van Westendorp).
• Launch shape the positioning and messaging, prepare the launch assets (website, demo script, pitch deck), design the channels and the funnel, all the way to launch day itself.
• Scale run a retrospective, stand up a CRM, growth sprints, growth loops, content, and sales, out to ABM (account-based marketing, targeting large accounts one by one) and partnerships.

What sets it apart from an ordinary box of prompts is that no skill floats free. Each one leans on frameworks with real names and real origins: positioning in the April Dunford style, pricing the Van Westendorp way, the OPE canvas from Wes Bush, the value proposition canvas from Strategyzer, each with step-by-step instructions, not just "try doing some positioning."

Part 2The design that matters more than the frameworks

Look only at the list of frameworks and it reads as one good PM/GTM set among many. But what really made me stop was the way it wires the work together. The author says it plainly in the README.

"Most AI prompts handle one task in isolation. But GTM doesn't work that way. Your competitor analysis should feed your positioning, your positioning should feed your pitch deck, your pitch deck should feed your sales process."

And it closes on the one line that holds the whole thing together: "Nothing exists in isolation." It does two things an ordinary box of prompts does not.

One, set the context once. There is a file called my-gtm-context.md where you fill in your product and market details in one place, and every skill reads it before it starts. You never have to retell your business each time you call a tool. As the README puts it: "You set up your product context once, then ask for what you need."

Two, each phase's output feeds the next. Every result lands in an outputs/ folder, and the next phase picks up the earlier work instead of starting from zero every time. The rules in its config file are explicit: read the context first, build on prior work, and cite the framework you used, do not answer in generalities.

The result is that instead of a set of disconnected documents, you get a playbook that hangs together. The positioning that comes out knows the competitors you analyzed earlier. The pitch deck that comes out speaks the same language as the positioning. That is the part where AI actually helps, not just being good at one piece at a time.

Part 3Lessons you can use even without GTM

Even if you are not planning a go-to-market right now, these three design moves are worth stealing for any AI work at all.

• Give the agent a context file it always reads first. Instead of retelling the same story every time, keep the context in one file and have the tool read it before it acts. It saves time and keeps answers from drifting.
• Make outputs cascade, not sit in separate blocks. Design it so earlier work is the raw material for later work. That is how complex work comes out coherent, instead of good piece by piece but impossible to assemble.
• Validate before you scale, don't guess for the customer. The skill files are firm about it: do not fill in answers from your own guesses, use the context as a starting point and then ask the user for the real detail. And there is a rule that a "good enough" deliverable shipped today beats a perfect one next month.

How to start

If you want to actually try it, you don't need all 12 phases.

1. Open Phase 1 (GTM Foundations) and fill my-gtm-context.md with your real product details.
2. Have it draft the first-phase work, say a value proposition or a 90-day plan, and watch whether it actually pulls in the context you filled.
3. When you move to phase two, notice that it carries the phase-one result forward rather than starting over. That is exactly where this design differs from ordinary prompts.
4. The decisions, which market to win first, what to charge, are still yours. The tool only speeds up the doing, it does not make the call.

gtm-strategist-skills is a good example of the bigger trend I have been watching: consulting work that used to cost thousands to tens of thousands is now a skill you can call for nearly free (and the skill set is itself a doorway to Maja Voje's book and course). But the thing I want you to take from this post is not whether it can replace a consultant. It is the design: set the context once, and make everything connect. This is the first in a series where I go through business-facing repos like this one, with more to come.