steadydata

For agents

Every dataset as an MCP tool

Two ways in. One is a URL and takes a minute. The other is a single file you run yourself, with nothing between you and the data.

The quick way: no install

Apify hosts an MCP server that exposes public store Actors as tools, and all 43 of our datasets qualify. You point your client at a URL, list the datasets you want, and sign in with your Apify account on first connection. Nothing to download and nothing of ours to trust.

mcp config
{
  "mcpServers": {
    "steadydata": {
      "url": "https://mcp.apify.com?tools=steadydata/app-store-app-details,steadydata/app-store-search,steadydata/company-career-site-jobs"
    }
  }
}

# add any of our datasets to the tools list, comma separated
# first connection opens a browser to sign in to Apify

Apify's server allows up to 30 requests per second and excludes full-permission and rental Actors. None of ours fall in those categories.

The self-contained way

If you would rather not have a hosted server in the path, run ours. One file, standard library only, Python 3.9 or newer. It talks to Apify with your token, on your machine. We never see it.

terminal
curl -O https://steadydata.io/steadydata_mcp.py
export APIFY_TOKEN=apify_api_...
python3 steadydata_mcp.py

Three optional settings. STEADYDATA_MAX_ROWS caps how many rows one call returns, 100 by default. STEADYDATA_ACTORS takes a comma separated list if you want a smaller toolbox. STEADYDATA_WAIT_S is how long it waits for a run, 120 seconds by default.

Either way, the price is in the tool

Each dataset carries its price per 1,000 rows in the tool description, so an agent can weigh a call before it makes it. You are billed by Apify for what you run. There is no account with us, because there is nothing to sign up for.