Roadmap
The MCP server is the first piece of a builder toolkit, not a submit endpoint. Here is what lands next.
Submitting a product is the smallest part of launching one. The work around it, finding the people who should hear about it, planning where else to post, watching what the launch did, keeping the listing current, is where the time actually goes, and it is all work an agent can do with the right tools.
So the MCP server is not a submit endpoint with extra steps. It is the place those tools land as they ship.
Shipped
search_products and get_product, anonymous.
The preflight audit, with the fix for every failing check.
submit_product, bound to your account by one browser click.
get_my_launches, including drafts waiting on you.
Coming next
First Users: an ICP audience report from a URL
Point it at a product and get back who the first hundred users should be: the segments, where they already gather, and the language they use about the problem. This exists today as a browser tool at /tools/find-users/. Exposing it as a tool means an agent can go from "here is my product" to "here is who to talk to and where" without a tab.
Directory campaign planner
list_launch_directories returns the catalog. The planner
turns it into an ordered plan for one product: which directories fit the category, what each one
requires, what to submit in which order, and what is already done. The requirements differ per
directory, and that is exactly the kind of bookkeeping worth handing to an agent.
Launch analytics
Impressions, upvotes over time, referrers and click through for your own launches, readable by an agent. Enough to answer "did the launch work" and "what should I do differently next time" with numbers rather than a feeling.
Badge management
The backlink badge is the one thing a free launch depends on, and today verifying it means waiting
for a crawl. A tool to check the current badge status of your listings, get the right snippet for
each, and trigger a re verification after you install one, closes the loop that
check_launch_readiness opens.
Changelog publishing
Listings carry a changelog. Publishing an entry from the agent that just shipped the release, rather than from a form a week later, is the version of this that actually happens.
What will not change
A human still finishes every launch.
Whatever else lands here, no tool will publish a launch, pick a plan, pick a week, or spend money. An agent prepares; a person decides at /submit/plan/; an admin reviews. Adding tools widens what can be prepared, never what can be published unattended.
Two other commitments are worth stating: the error vocabulary on Errors stays
stable, and breaking changes arrive as a new path version rather than as a changed meaning under
/api/v1/.
Ask for something
If you are building against this and the missing piece is not on the list, say so. Mail
support@nicklaunches.com with what you are trying to automate. Tools that several people ask for
get built first.