The product-marketing skill builds one shared document, .agents/product-marketing.md, that holds the foundation for all marketing work: what the product is, who it serves, how it differs, how customers talk and how the brand sounds.
It can draft a first version from the repository, or walk through each section in conversation. Every later update bumps a document version and adds a changelog line, and a Learnings table records what shipped work taught the team.
What You Get
- A .agents/product-marketing.md file with fourteen sections from product overview to learnings
- A first draft built from the README, landing pages and copy in the repository
- Personas, objections, switching dynamics and customer language captured in a fixed layout
- A versioned changelog that shows how positioning changed and why
- A Learnings table of tested lessons that drafting skills read before writing
Common Use Cases
- Starting a new project before running any other marketing skill
- Updating the file after a repositioning or a change of target customer
- Checking whether positioning and ideal customer are clear enough to write copy
- Recording a test result so later copy and ad drafts respect it
What Makes It Unique
Other marketing skills in the set all read this file first, so it works as shared memory for the whole library. It pushes for verbatim customer phrases, includes the four forces of switching (push, pull, habit, anxiety), and treats the document as versioned with a changelog and a retire-not-delete rule for learnings.
Under The Hood
The skill is an instruction file plus one reference on extracting brand voice. It has no scripts. In auto-draft mode the agent reads the repository’s README, landing pages, copy, package manifest and docs, and it writes or moves the context file under the project’s .agents folder. It does not fetch web pages, call APIs or run commands.