How do I connect App Store reviews to Cursor?

Add the ReplyArgus MCP server to Cursor. Create a free ReplyArgus account, connect App Store Connect or Google Play, copy your MCP endpoint from the agentic-tools settings, and paste it into Cursor's mcp.json (or Settings → MCP → Add new server). Cursor authorizes over OAuth, and the review tools appear in the Agent.

Where does the ReplyArgus MCP config go in Cursor?

In an mcpServers block. A global config lives at ~/.cursor/mcp.json; a project-scoped one goes in .cursor/mcp.json at your repo root. Add a "replyargus" entry with a "url" field set to the endpoint from your ReplyArgus agentic-tools settings, then reload Cursor.

Will Cursor publish replies to my store automatically?

No, not by default. Every reply the Agent drafts lands in your ReplyArgus approval queue. A reply only posts to a real store when you approve it, or under an auto-publish rule you deliberately set inside ReplyArgus first by rating, keyword, or language.

Does the same server work in Claude and ChatGPT too?

Yes. Because it's built on MCP, an open standard, the same ReplyArgus server answers Claude, Claude Code, ChatGPT, and Cursor. You set it up once and use it from whichever assistant you prefer — the review tools are identical across all of them.

Do I need a paid plan to use the connector in Cursor?

You can start on the free plan (one app, 100 replies a month, manual approval), and the connector works from day one (it's gated by role, not tier). Heavier automation, more apps, the auto-ingested knowledge base, and the roadmap export sit on the paid tiers starting at $29/month.

Who on my team can add the connector?

Only Owner and Admin roles. The server can read every review and queue replies across your apps, so members, drafters, and viewers can't add it. An owner or admin authorizes it once for the workspace, and each teammate pastes their own endpoint into their Cursor config.

All posts

ProductJul 8, 2026 · 7 min

App Store Reviews in Cursor: Set Up the ReplyArgus MCP Server

Wire the ReplyArgus MCP server into Cursor and run your App Store + Google Play review queue without leaving your editor. Config, prompts, and dev workflows.

The Argus Team

Reply Argus

You live in Cursor all day. Your App Store and Google Play reviews live in a browser tab you keep forgetting to open. Connect one MCP server and that gap closes: you fix the crash in your editor, then type into the same chat panel, "pull the 1-star reviews about this crash and draft replies telling everyone it's patched in 4.3." Cursor reads your real review inbox, drafts grounded replies in each reviewer's language, and queues them for approval. No tab-switch, no context-switch.

Almost nobody wires reviews into their editor yet, so there's no worn path to copy. This is the whole setup, start to finish: the exact `mcp.json` config, the prompts that carry the daily grind, and why having reviews sit next to the code that caused them changes how you triage.

What connecting reviews to Cursor actually means

Cursor's Agent can read your repo and write code, but it has no login to App Store Connect and no key into the Google Play Console. Out of the box it can't see a single one of your reviews. MCP (the Model Context Protocol, an open standard Anthropic introduced in late 2024) is the bridge that fixes that. It's a universal adapter between an AI assistant and a live external system, and it's the same mechanism behind the connectors Cursor already supports for GitHub, Linear, and the rest. If you want the deeper mechanics of how a review inbox gets exposed as callable tools, we lay it out in [app reviews over MCP](/blog/app-reviews-mcp).

ReplyArgus ships an MCP server that re-exposes your review inbox as named tools the Cursor Agent can call: list unanswered reviews, draft a grounded reply, break a spike down by theme, queue it for approval. The model stops guessing from training data and starts operating on your real, current queue. The store stays the source of truth; Cursor just becomes another surface you drive it from. The same server also answers Claude and ChatGPT, so the setup carries. If your team lives in the terminal instead, the [Claude Code walkthrough](/blog/replyargus-mcp-in-claude-code) is the same connector, a different door.

Connect the ReplyArgus MCP server to Cursor

Two moves: set up ReplyArgus so it can see your reviews, then point Cursor at it. Under ten minutes, most of which is the store connection you'd do anyway.

1
Step 1 — Create an account and connect a store
Sign up free, add your app, and connect App Store Connect or Google Play so ReplyArgus can read your reviews. This is the ordinary onboarding, nothing Cursor-specific yet.
2
Step 2 — Copy your MCP endpoint
In ReplyArgus, open the agentic-tools settings and copy your server URL. That's the address Cursor will connect to, scoped to your workspace.
3
Step 3 — Add it to Cursor's mcp.json
Open Cursor Settings → MCP → Add new MCP server, or edit the config file directly (shown below). Paste your endpoint as the server URL and save. Cursor pops a browser window to authorize (a standard OAuth handshake), then shows a green dot when the server is live.
4
Step 4 — Confirm the tools loaded
In an Agent chat, ask "what review tools do you have?" Cursor should list the ReplyArgus tools. Then run "summarize my unanswered 1-star reviews from the last 7 days." A live themed digest means you're connected.

json

{
  "mcpServers": {
    "replyargus": {
      "url": "PASTE_YOUR_MCP_ENDPOINT_HERE"
    }
  }
}

Global config lives at ~/.cursor/mcp.json; a project-scoped copy goes in .cursor/mcp.json at your repo root. Replace the url value with the endpoint from your ReplyArgus agentic-tools settings, save, and Cursor picks it up on reload.

Who can connect it

The MCP connector is scoped to Owner and Admin roles only. That's deliberate: a server that can read every review and queue replies across your apps isn't something a view-only teammate should hold. On a team, an owner authorizes it once, and a downgraded account loses access on its next call. A project-scoped .cursor/mcp.json is handy here: commit it (without the live endpoint) so teammates know the server exists, and each authorized person pastes their own URL.

The workflows that pay off when reviews live in your editor

A dashboard makes you go to the reviews. Cursor brings them to where you already are, which changes what you reach for them for. The Agent can hold your codebase and your review inbox in one context, so the reviews stop being a support chore and start being debugging signal. A few flows that only make sense inside the editor:

Fix, then answer, in one sitting — you just shipped a patch for a logout bug. Ask the Agent to pull every review that mentions it and draft replies saying it's fixed in the next build. The users who flagged the bug hear back the same afternoon you closed it.
Turn complaints into a repro — "cluster the 1-star reviews since 3.4.1 and quote the exact symptoms users describe." That's a bug report with evidence attached, written by real users, dropped straight into the buffer next to the file you're editing.
Reviews as release-note fodder — writing the changelog? Ask which themes drove the last two weeks of complaints, and phrase your "what's fixed" notes around what people actually asked for.
Triage without leaving flow — "summarize this week's angry reviews by theme, ranked by count." You get the shape of the fire in ten seconds without opening a browser or breaking concentration.

You don't need clever prompt engineering for any of this. Because each tool is named and typed, plain requests route to the right call — ask for an outcome and the Agent chains `list_reviews` into `theme_breakdown` into `draft_reply` without you spelling out the plumbing. Three prompts cover most of a review week:

text

1. Triage
"Summarize my 1- and 2-star reviews from the past week
 for [APP], grouped by theme, ranked by how many
 reviews mention each."

2. Draft (grounded, no publish)
"Draft replies to every unanswered review in [APP], each
 in the language the review was written in. Keep Google
 Play replies under 350 characters. Don't publish —
 queue them for me to approve."

3. Root-cause next to the code
"Which single bug is driving the most 1-star reviews
 right now? Cluster the reviews that mention it, quote
 the symptoms, and show me the count."

Fill in your app name and go. The first triages, the second drafts, the third turns a review spike into something you can actually debug.

Prompt 2 is where the time goes back in your pocket, so it's worth being precise about what it does. The Agent isn't inventing a voice or a feature set. `draft_reply` anchors each reply to your knowledge base and your history of approved replies, in the reviewer's own language, across 100+ languages both directions. Matching an incoming review's language is a discipline we go deep on in [replying to app reviews in any language](/blog/reply-to-app-reviews-in-any-language). One accuracy detail the tools handle for you: Google Play caps developer replies at a hard 350 characters, while Apple publishes no official limit (community testing suggests a few thousand characters), so drafts are sized to the store they're bound for. Here's the kind of grounded draft it hands back for a furious one-star:

Update 3.4.1 signs me out every time I close the app. Retyped my password five times today. This is broken.

Getting signed out every launch is not okay, and I'm sorry 3.4.1 did that. It's a confirmed session bug and the fix is already in review. Until it ships, backgrounding the app instead of force-quitting keeps you logged in, and if you email support@app.com I'll flag your account the moment the patch is live. — Sam, Northwind team

It names the bug, offers a real workaround, and promises no date it can't keep. That restraint is exactly what grounding buys you. A reply pulled from your own knowledge base reads like your team, not a form letter, and that's the difference between one that nudges a rating up and one that gets ignored. If you want the full move on the angry ones, we break it down in [how to respond to negative app reviews](/blog/how-to-respond-to-negative-app-reviews).

The Agent drafts — it does not silently publish

Every reply Cursor produces lands in your ReplyArgus approval queue by default. `publish_reply` only fires on a reply you've approved, or under an auto-publish rule you deliberately set inside ReplyArgus first (by rating, keyword, or language). Nothing goes live to a real store from a chat message on its own, and Cursor asks before running a tool anyway, unless you've turned that off. If you're weighing whether to let anything post unattended, read our honest take on [auto-publishing review replies](/blog/is-it-safe-to-auto-publish-app-review-replies) before flipping that switch.

Reviews next to the code that caused them

The root-cause prompt is where this stops being a faster reply tool and becomes an actual workflow. Ask the Agent to cluster the reviews behind a spike and `theme_breakdown` gives engineering what it needs: "9 reviews since 3.4.1 mention forced logout, here are the quotes." You're already in the file where that lives. The distance between the complaint and the fix collapses to zero.

That signal doesn't get stranded in a chat log, either. ReplyArgus clusters reviews into a PM roadmap board that exports to Jira, Notion, Google Sheets, or DevRev, so the theme the Agent surfaces becomes a tracked item you can push wherever your team plans work. The connector does the reading and the drafting; the board turns the pattern into a ticket. And the loop is worth closing fast — Google's own I/O 2019 data showed that apps whose developers reply gain about +0.7 stars on average, so [replying to reviews genuinely moves the number](/blog/does-replying-to-app-reviews-raise-your-rating), and a connector makes that habit cost you a sentence instead of an afternoon.

Start free — draft your first reply from Cursor in minutes

Spin up a free ReplyArgus account, connect a store, paste the endpoint from your [agentic-tools settings](/agentic-tools) into Cursor's mcp.json, and ask the Agent to draft its first reply. Free plan, no card required: [start free](/signup).

Frequently asked

How do I connect App Store reviews to Cursor?: Add the ReplyArgus MCP server to Cursor. Create a free ReplyArgus account, connect App Store Connect or Google Play, copy your MCP endpoint from the agentic-tools settings, and paste it into Cursor's mcp.json (or Settings → MCP → Add new server). Cursor authorizes over OAuth, and the review tools appear in the Agent.
Where does the ReplyArgus MCP config go in Cursor?: In an mcpServers block. A global config lives at ~/.cursor/mcp.json; a project-scoped one goes in .cursor/mcp.json at your repo root. Add a "replyargus" entry with a "url" field set to the endpoint from your ReplyArgus agentic-tools settings, then reload Cursor.
Will Cursor publish replies to my store automatically?: No, not by default. Every reply the Agent drafts lands in your ReplyArgus approval queue. A reply only posts to a real store when you approve it, or under an auto-publish rule you deliberately set inside ReplyArgus first by rating, keyword, or language.
Does the same server work in Claude and ChatGPT too?: Yes. Because it's built on MCP, an open standard, the same ReplyArgus server answers Claude, Claude Code, ChatGPT, and Cursor. You set it up once and use it from whichever assistant you prefer — the review tools are identical across all of them.
Do I need a paid plan to use the connector in Cursor?: You can start on the free plan (one app, 100 replies a month, manual approval), and the connector works from day one (it's gated by role, not tier). Heavier automation, more apps, the auto-ingested knowledge base, and the roadmap export sit on the paid tiers starting at $29/month.
Who on my team can add the connector?: Only Owner and Admin roles. The server can read every review and queue replies across your apps, so members, drafters, and viewers can't add it. An owner or admin authorizes it once for the workspace, and each teammate pastes their own endpoint into their Cursor config.

Try it

Let Argus draft your next reply.

Watch it answer a real review in your voice. 10-day trial, no card to begin.

Start free

See the features or pricing.

Keep reading

Product7 min

Manage Your App Store Reviews From Claude: The 5-Minute MCP Setup

Connect one MCP server and run your App Store + Google Play review queue from a Claude chat: summarize complaints, draft grounded replies, queue for approval.

Product8 min

The MCP Server for App Reviews: Connect Your AI Agent to Your Store Reviews

An app reviews MCP server hands your live App Store and Google Play reviews to Claude, ChatGPT, or Cursor as callable tools. Here's what it exposes and how to wire it up.