Storelift

StoreliftFeatures › MCP server

AI layer

MCP server

The dashboard is one way to read your measurements. The other is to ask for them in the assistant you already have open. The measurement states travel with the data — an assistant reading our API can tell "not in the list" from "not measured", which is exactly the distinction a model would otherwise paper over.

Start free →Updated:

In measured numbers

8tools the server exposes
0of them can write anything
2.1OAuth on the remote server — no key to paste
3measurement states that survive the trip

Every figure on this page is read from a code path, not written by hand — the same discipline the product applies to your own data.

What the tools return

ToolWhat it answers
list_appsWhich apps this account measures
get_keywordsCurrent rank per keyword and storefront, with measurement state
get_historyThe stored series for a keyword, gaps included
get_rivalsThe apps ranked above you, and their ratings
get_store_pageThe current listing as the store serves it
get_reviewsRecent reviews, including the crash-shaped ones
get_chartsTop-chart positions for a category and storefront
get_ai_visibilityWhether assistants name the app, per engine

Two ways to connect

  1. Remote (recommended). A hosted Streamable HTTP server with OAuth 2.1 — you authorise it once from the assistant, and there is no key to paste or rotate.
  2. Local (stdio). The storelift-mcp package on npm, run next to the assistant with an API key from your account.

The remote server also ships prompts — ready-made questions like "which keywords lost ground this week" — so the assistant does not have to be told how to assemble the query.

What you do with it

Ask in plain language

“Which of my keywords lost ground in Germany this week?” is a question, not a dashboard filter.

Set the server up

Put ASO next to your other data

An assistant that already sees your analytics can now see your ranks in the same conversation.

See where the numbers come from

Draft with evidence in context

Metadata written next to the actual ranks is a different draft from metadata written from memory.

Compare the ASO MCP servers

What this does not do

What an assistant cannot do here

Every tool is read-only. There is no write path at all.

The measurement states travel with the data, which is the part a model would otherwise paper over:

  • It does not write anything. No path from an assistant to your listing, your keywords or your billing.
  • It does not measure on demand. The tools read what the nightly run collected; they do not trigger new searches against the stores.
  • It does not flatten the states. An unmeasured day says so, rather than returning a zero the model would then reason about.

What each plan includes

FreeProStudio
API access (required for the local server)YesYes
Remote MCP serverYesYes
Keywords the tools can read25250500
History the tools can read30 days365 days365 days

Pro is $20/month ($18 billed yearly), Studio $60/month ($54 billed yearly).

Frequently asked

How many tools does the server expose?

8: list_apps, get_keywords, get_history, get_rivals, get_store_page, get_reviews, get_charts and get_ai_visibility.

Can an assistant change my keywords or metadata through it?

No. Every tool is read-only, and there is no write path from MCP to your listing, your tracked keywords or your account.

Do I need to paste an API key?

Not for the remote server — it uses OAuth 2.1, so you authorise it once from inside the assistant. The local stdio package does use an API key from your account.

Which plans include it?

Pro and Studio. The free plan does not include API access.

See this for your own app

Paste your store link, add your keywords. Rank, the apps above you and demand are measured every day.

Start free →

Read next