Nalya is under construction — things may move, change or break.
Nalya/Blog
About
MCP·6 min read

Manage your Play Store app from Claude Code or Cursor with MCP

If you build with Claude Code, Cursor, or any other agentic editor, you already have a place where you read, decide, and act all day. The Model Context Protocol (MCP) is what lets that place see beyond your repo - and one of the most useful things to point it at is your own app's Play Store presence.

MCP in one paragraph

MCP is an open protocol that lets AI assistants connect to external tools and data sources through a standard interface. An MCP server exposes capabilities - "list recent reviews," "check rollout status," "draft a reply" - and any MCP-capable client (Claude Code, Cursor, Windsurf, Zed, and a growing list) can discover and call them. One server, every editor.

What this looks like for a Play app

Nalya ships an MCP server that speaks for your connected Play account. Once added, your coding agent can answer questions and take actions that used to mean opening Play Console in a browser tab:

  • "Any new reviews since yesterday? Summarize the complaints" - and then "draft replies to the negative ones."
  • "Where is the 2.4.1 rollout, and how do its vitals compare to 2.4.0?" - answered live from your tracks, no browser tab.
  • "Which locales convert worst on the store listing?" - followed by a drafted short-description test for the weakest one.
  • "How did revenue move after the price change on the annual plan?"

The workflow this unlocks is the interesting part: you fix a crash in your editor, and in the same session you check which reviews mentioned it, draft replies telling those users it's fixed, and confirm where the rollout stands. Code and operations stop being separate rooms.

Reads are open; writes are gated

An agent with write access to your production app deserves suspicion, so the write path is deliberately conservative: reads are always available, and writes go through the same supervised approval loop as the rest of Nalya. A reply drafted from your editor is still a draft until it's approved; a listing change still waits for your go-ahead before it touches Play. Autonomy is something you grant per capability, not something a tool call assumes.

Why the editor is the right surface

Dashboards assume you'll come to them; you mostly don't. Your editor is where you already are, and agentic editors are increasingly where the whole dev loop happens. Putting app operations behind MCP means the marginal cost of checking on your app drops to one sentence - which is the difference between operational work that happens and operational work that piles up (the long tail every solo dev knows).

And if you'd rather not drive it by hand at all, everything the MCP server exposes is the same machinery Nalya runs herself - the editor is just one more seat at the same table.