Describe your product once. From then on your AI knows what you are building, which of the 36 directories will take it, what is missing for the ones that will not, and how each form wants to be filled. Ask it to submit and it does the typing, in your browser, and writes down what happened.
The first four steps are the setup, once, on any MCP client. The three after it are just you talking to your agent. The directory itself stays free and needs none of this.
In the dashboard, open Connect your agent and create a token. It is shown once, because only a hash of it is stored, so copy it before you close the page.
Connect your agentPick your client below and follow its three lines. Anything that speaks MCP over HTTP will work; the token goes in an Authorization header.
claude mcp add --transport http submitmap \ https://submitmap.com/api/mcp \ --header "Authorization: Bearer smap_your_token_here"
Add --scope user to make it available in every project rather than this one. Drop the --header line to connect without an account and keep the three public tools.
The same steps work on claude.ai in a browser; it is one setting shared by the app and the site. That dialog takes a URL and OAuth credentials, with nowhere to put a token, so a connector added here gets the tools that need no account: search the directory, read a platform, and qualify a product you describe in the conversation. To store a project and let it submit, use one of the other clients, which can send the token.
{
"mcpServers": {
"submitmap": {
"type": "http",
"url": "https://submitmap.com/api/mcp",
"headers": {
"Authorization": "Bearer smap_your_token_here"
}
}
}
}Cursor picks the file up without a restart. The green dot next to the server name means it connected.
{
"servers": {
"submitmap": {
"type": "http",
"url": "https://submitmap.com/api/mcp",
"headers": {
"Authorization": "Bearer smap_your_token_here"
}
}
}
}{
"mcpServers": {
"submitmap": {
"type": "http",
"url": "https://submitmap.com/api/mcp",
"headers": {
"Authorization": "Bearer smap_your_token_here"
}
}
}
}Some clients call this transport streamable-http rather than http; the two names mean the same thing here. If yours has a UI for adding a remote server, the URL alone is enough to get the public tools.
Start the session in your product's folder and ask. Your agent already has the README, the package metadata, the site copy and whatever is in your assets folder, so it can fill the project in from what it can see rather than from what you retype. Describe it yourself if you would rather, or use the dashboard form.
"Add my product to SubmitMap."
From inside the product that is the whole prompt. Somewhere else, or when the repository is not the product, say it instead:
"Add my product to SubmitMap. It is called {NAME}, it lives at
{URL}, it is a {WHAT IT DOES} for {WHO}, pricing is {FREE /
FREEMIUM / PAID}, and it is {LIVE / PRE-LAUNCH}."Every platform is sorted into ready to submit, reachable once you supply what is missing, and out of reach, with the exact rule behind each answer.
Pick a platform and your agent gets the runbook: what is still missing, the values for each field, the steps and the gotchas. It fills the form in your own browser and records what happened.
It fills the forms, in your browser, while you watch. That is deliberate: the automation runs where you are already signed in, so nothing about your accounts has to be handed to us.
No. When a platform asks you to sign in, the agent stops and hands the keyboard back to you. It will not type credentials and it will not create accounts on your behalf.
The listing URL, when it was submitted, and when it goes live, on the submission it just made. That is what turns "did I ever submit to that one?" into an answer.
No. The directory, the guides and the route are free and need no account, and your agent can search and qualify a product without a token at all. The account exists for storing a project and tracking what you sent.
Building the agent rather than using one? The tools and the dataset are documented here.