Install one MCP server and your assistant can answer “how much does this app make?” with a real estimate and its confidence label — no dashboards, no scraping.
Create an Appdex account and copy your key from the account page. Free tier: 50 requests/day.
Claude Desktop (claude_desktop_config.json) or Cursor (.cursor/mcp.json):
{
"mcpServers": {
"appdex": {
"command": "npx",
"args": ["-y", "appdex-mcp"],
"env": { "APPDEX_API_KEY": "adx_your_key_here" }
}
}
}Restart the host afterwards. The package is fetched by npx; no global install needed.
“What does Calm earn per month on iOS?” — the agent calls estimate_app and answers with a range plus its confidence label.
Prefer plain HTTP? Any agent that can make a request can use the same endpoint:
curl -H "Authorization: Bearer $APPDEX_API_KEY" \ "https://app-dex.com/api/v1/app?id=570060128"
{
"name": "Duolingo: Language Lessons",
"category": "Education",
"monthlyRevenue": { "range": "$48.0M – $48.0M", "confidence": "calibrated" },
"monthlyDownloads": { "range": "5.0M – 5.0M" },
"url": "https://app-dex.com/a/duolingo-language-lessons-570060128"
}| Tool | What it returns |
|---|---|
estimate_app | Monthly revenue + downloads for one iOS app, with a confidence label. Accepts a name or an App Store id. |
search_apps | Resolves a name to App Store ids (no figures) — feeds estimate_app. |
| Free | Paid | |
|---|---|---|
| Requests / day | 50 | 5,000 |
| Revenue & downloads | ranges | ranges + point estimates |
| Confidence label | yes | yes |
| US chart rank · revenue-per-install | — | yes |
iOS App Store only · figures are gross · recomputed nightly · single-app lookups (no bulk export).
An MCP (Model Context Protocol) server that lets an AI agent query iOS App Store revenue and download estimates. It is published on npm as appdex-mcp and works with Claude Desktop, Cursor, and any MCP-capable host.
Yes. A free key from your Appdex account gives 50 requests per day and returns ranges with a confidence label. Paid plans raise the limit to 5,000/day and add point estimates, US chart rank and revenue-per-install.
estimate_app returns monthly revenue and downloads for one app with its confidence label. search_apps resolves an app name to App Store ids so the agent can chain the two.
Every figure ships with a confidence label: verified (reported data), calibrated (modeled but calibrated against verified panel data) or modeled (limited public signal). Read the range, not the midpoint.
No. The API is single-app lookup by design. There is no bulk or CSV endpoint.