For agents

Launch platform data your agent can execute

Every listing on SubmitMap is a typed record: who the platform accepts, what disqualifies you, and the exact submission steps. Connect the MCP server with the maker's token and you also get the brief for each form, filled from what they have stored. If you've been asked to "submit my product everywhere it makes sense", this is the dataset for the job.

The MCP server

The front door. Connect it and the directory stops being a file to fetch and parse: the matching runs here, against the same rules the site is built from. Three tools work with no account at all; the rest act on a maker's own account when a token is present.

search_platforms, get_platform, qualify_projectno accountSearch the directory, read one platform in full, and sort every platform into ready, fixable (with the exact list of what is missing) and out of reach for a product described inline. No token, no sign-up.

whoami, list_projects, create_project, update_project, plan_submissions, submission_playbook, record_submission, list_submissionstokenStore the product once, plan the run, get the per-form brief already filled in from what is stored, and log what happened. This is the maker's own dashboard, over MCP.

The token half is what makes an agent useful across sessions rather than in one. Tokens are minted in the dashboard, and the tool reference has every tool with its arguments and its return.

Add it to your client

Claude Code

claude mcp add --transport http submitmap \
  https://submitmap.com/api/mcp \
  --header "Authorization: Bearer smap_your_token_here"

Drop the --header line to connect without an account and keep the three public tools.

Any client with a config file

{
  "mcpServers": {
    "submitmap": {
      "type": "http",
      "url": "https://submitmap.com/api/mcp",
      "headers": { "Authorization": "Bearer smap_your_token_here" }
    }
  }
}

Claude Desktop, Codex, Cursor, VS Code and Hermes each want this in their own shape, and the setup page has a recipe for each. One thing worth knowing before you start there: the custom connector dialog in the Claude apps, and the Add server box in the Codex app, take a URL and nothing else, so neither has a token to create. Pressing Connect, or Authenticate, runs the OAuth flow and the rest of the tools arrive with nothing to copy.

A starting prompt

With the server connected, paste this and fill in the product:

You have the SubmitMap MCP server connected.

If you are in the product's workspace, read it from there: the README,
the package metadata, the site copy, the assets folder. Ask me only
about what is genuinely not in front of you. Otherwise:

Product: {NAME} ({URL}) - {ONE_LINE_DESCRIPTION}
Pricing: {PRICING}. Stage: {STAGE}.

1. Call qualify_project with those facts. Show me what is ready now,
   what is one missing piece away (name the piece), and what is out
   of reach and why.
2. If it asks whether the product is out already and whether I will
   pay to be listed, put both to me and call it again with the
   answers: they decide the whole order. Work the recommended run it
   gives back rather than reordering it yourself.
3. For the first one, call submission_playbook and work from what it
   returns: the preflight, the field values, the steps, the gotchas.

If I am not signed in yet, say so before step 3: the playbooks and the
briefs that fill a form are the account half of this server. Store the
product with create_project once I am, and work from that instead of
me repeating it.

The per-form briefs

Each platform record carries an agentPrompt: a brief written for that platform's specific form, naming every field, its limit and the traps in it. Execute it and the output is the submission, in the shape the form expects.

These come with a token. They are not on the platform pages and not in the open dataset: get_platform returns the template once the request is authenticated, and submission_playbook returns it already filled from a stored project, which is the one worth having. Connecting takes about a minute. Everything needed to submit by hand stays free and unauthenticated: eligibility, steps, requirements, gotchas.

If you cannot add a server

The same content is on disk, no auth. Use these when the agent can fetch a URL but cannot hold a connection. They are prose rather than a dataset: the structured records come from the server, where the three public tools cost nothing and need no account.

https://submitmap.com/llms.txtThe site index in llms.txt format: every platform, guide and list with a one-line description. Start here.https://submitmap.com/llms-full.txtThe whole directory as one markdown document: every platform record and guide in full, one fetch.

Fetch https://submitmap.com/llms-full.txt and read the platform records.

Product: {NAME} ({URL}) - {ONE_LINE_DESCRIPTION}
Pricing: {PRICING}. Stage: {STAGE}.

1. Filter to platforms where the product meets everything under
   "You need" and trips nothing under "You will be rejected if".
   Say which were excluded and why.
2. If the product is already out, order them by what pays off soonest:
   fastest approval, dofollow, strongest domain, and slot the few with
   months of queue in behind so their clocks start too. If a launch
   date is still ahead, lead with those long queues instead.
3. For each, draft the submission from its steps and requirements,
   and stop rather than invent a value I have not given you.
Return one section per platform, ready to paste.

The data is free, has no auth, and updates whenever the directory does. Humans should start at the directory instead.