Potencialize seu agente
O comando do próprio publicador. Não é um endpoint do PlugMCP.
Host a copy with our keys and logs.
npx -y [email protected]O Claude Code não tem link de instalação para um comando local. Copie o comando abaixo e execute-o em um terminal.
claude mcp add appstore-api-mcp -- npx -y appstore-api-mcp@1.16.4Execute em um terminal.
{ "mcpServers": { "appstore-api-mcp": { "command": "npx", "args": [ "-y", "[email protected]" ] } }}claude_desktop_config.json
URL remota ou um comando na sua máquina? Como as duas instalações diferem
Rodar no PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
AbrirDo publicador
Run App Store Connect by talking to your AI. An MCP server that turns Apple's App Store Connect API into plain-language actions — edit your listings, , audit your whole portfolio, and reach the entire API, from whatever AI agent you already use.
MCP server for Apple App Store Connect — edit listings (keywords, descriptions, titles, screenshots), track analytics (downloads, proceeds, subscriptions, retention), run a fleet-wide ASO audit, preview changes with dry-run, and reach the full API. Works
🚀 Audit your whole portfolio in one shot. "Audit all my apps for ASO gaps" → a ranked report across every app: empty keyword fields, missing subtitles, under-used 100‑char keyword space, single‑locale listings, missing screenshots. Built for indies shipping dozens of apps — not one at a time.
🛡️ See every change before it ships. Dry-run mode returns the exact diff (old → new) with Apple's character limits pre-checked, so nothing hits your live listing by surprise.
📊 Track performance, not just listings. Pull downloads, proceeds, subscriptions & retention, and engagement straight from the Sales, Finance, and Analytics report APIs — parsed into rows, gzip handled for you.
🌐 The whole API, not a curated slice. First-class tools for the daily work (keywords, descriptions, titles, subtitles, screenshots, versions) plus a
raw_requestescape hatch for everything else — TestFlight, pricing, in-app purchases, customer reviews (read & reply), and more.
🔒 Safe by default if you want it. Worried about a write-capable tool touching your live apps? Set
APPSTORE_MCP_READ_ONLY=trueand it physically cannot write — only reads, audits, and reports run. Or block just the risky bits (releases, pricing, public review replies). It's enforced by the server, so even a misbehaving agent can't get around it. → Safe mode
“Set the keywords for my budgeting app to budget, expenses, money tracker.” “Which of my apps are missing a subtitle or screenshots?” “Dry-run a shorter description for MyApp, then upload these 6.7″ screenshots.”
npx, no build) — or paste a prompt and let your agent set it up for youRequirements: Node.js ≥ 18 and an Apple Developer account with an
App Store Connect API key. Everything works from any
OS — except the optional build & ship tools (archive_app / upload_build),
which additionally need a Mac with Xcode.
Don't want to touch config files? Paste this into your coding agent (Claude, Codex, Cursor, Windsurf, Antigravity, Gemini CLI, …), fill in your three values, and it will configure your client and verify it:
Set up the "appstore-api-mcp" App Store Connect MCP server for me.
My App Store Connect API credentials:
- Key ID: <YOUR_KEY_ID>
- Issuer ID: <YOUR_ISSUER_ID>
- Path to my .p8 private key file: <ABSOLUTE_PATH_TO_AuthKey_XXXX.p8>
Please:
1. If any of the three values above are still placeholders or I didn't provide
them, do NOT stop with an error — guide me to get them, then wait for me:
- Key ID & Issuer ID: go to App Store Connect → Users and Access →
Integrations tab → "App Store Connect API". The Issuer ID is shown at the
top of that page; the Key ID is listed next to each key.
- No key yet? Click the + to generate one, give it the "App Manager" role
(enough to edit metadata/keywords/screenshots), then Download the .p8
file — you can only download it once, so save it somewhere safe.
- The .p8 path is the absolute path to wherever you saved that file.
2. Detect which MCP client I'm using and add the server to its config (run it
with: npx -y appstore-api-mcp).
3. Pass these env vars: ASC_KEY_ID, ASC_ISSUER_ID, and ASC_PRIVATE_KEY_PATH
(point ASC_PRIVATE_KEY_PATH at my .p8 path — reference the PATH, do not
inline the key contents).
4. Install it at user/global scope (for Claude Code use `--scope user`).
5. Do NOT print, echo, log, or commit the key. Keep the .p8 outside any git repo.
6. When done, verify by listing my App Store apps and report the result.
Config formats per client: https://github.com/fil-technology/appstore-api-mcp/blob/main/docs/CLIENTS.md
Give the file path, not the key contents — that keeps the private key off the chat transcript. More detail + a remote/cloud variant in docs/AGENT-SETUP.md.
Prefer to do it manually? See below.
This is a standard stdio MCP server, so every MCP client uses the same
command (npx -y appstore-api-mcp) and the same three ASC_* env vars. Most
clients (Cursor, Cline, Windsurf, Claude Desktop, Gemini CLI, Antigravity, …)
take this exact block — just put it in that client's config file:
{
"mcpServers": {
"appstore-api": {
"command": "npx",
"args": ["-y", "appstore-api-mcp"],
"env": {
"ASC_KEY_ID": "YOUR_KEY_ID",
"ASC_ISSUER_ID": "YOUR_ISSUER_ID",
"ASC_PRIVATE_KEY_PATH": "/absolute/path/to/AuthKey_XXXXXXXXXX.p8"
}
}
}
}
A few clients differ (VS Code uses servers, Codex uses TOML, Zed uses
context_servers) — see the Supported clients table and
docs/CLIENTS.md for each exact format/location.
Claude Code has a one-line CLI shortcut instead of editing a file:
# --scope user installs it for ALL your projects (recommended).
# Remove the --scope user line to install for the current project only.
claude mcp add appstore-api \
--scope user \
--env ASC_KEY_ID=YOUR_KEY_ID \
--env ASC_ISSUER_ID=YOUR_ISSUER_ID \
--env ASC_PRIVATE_KEY_PATH=/absolute/path/to/AuthKey_XXXXXXXXXX.p8 \
-- npx -y appstore-api-mcp
After configuring, restart the client and ask it to "list my App Store apps" to confirm it works.
This is a standard stdio MCP server — it works with any MCP-compatible agent.
The command (npx -y appstore-api-mcp) and the three ASC_* env vars are always
the same; only each client's config format/location differs.
Copy-paste config snippets for each are in docs/CLIENTS.md.
.p8 file — you can only download it once. Store it somewhere safe.The Issuer ID is at the top of the page; each key's Key ID is in its row. The
.p8is downloaded from the + / key actions. These map toASC_ISSUER_ID,ASC_KEY_ID, andASC_PRIVATE_KEY_PATH.
Full walkthrough with screenshots-worth of detail: docs/SETUP.md.
All configuration is via environment variables.
See .env.example for a copy-paste template.
You can run this read-only. Tool-level guardrails — enforced by the server, not "please ask the agent" — so even a misbehaving agent can't write. Recommended for shared machines, demos, or CI. Set any of these env vars to lock things down:
Blocked calls return a clear error explaining which flag to change. Run the
doctor tool to see the active mode and verify your whole setup.
Reports need a higher-privilege key. The sales/subscription/finance/analytics tools require an API key with the Admin, Finance, or Sales role — an App Manager key (fine for metadata/keywords/screenshots) returns
403for reports. Sales/finance also need your Vendor Number (App Store Connect → Payments and Financial Reports) viaASC_VENDOR_NUMBERor thevendorNumberargument. Full walkthrough: docs/ANALYTICS.md.
A compact set of high-level tools, plus raw_request for everything else.
Full parameter reference: docs/TOOLS.md.
⚙️ Rate-limit aware: the client automatically backs off and retries on App Store Connect's
429(hourly quota), so large sweeps (e.g. auditing 60+ apps) don't fall over.
🛡️ Dry-run: the
update_*tools acceptdryRun: true— they return a field-by-field diff (old → new) with length/limit checks and write nothing. Drop the flag to apply.
Once it's connected, just talk to your agent in plain language. A few things to try:
Listings & ASO
Performance & feedback
Release management
Pipeline & accounts
🛠️ Build & ship — additional, Mac-only (see note below)
Dedicated tools cover it all: app management, localization, screenshots (incl. seeing them), sales/subscriptions/analytics, ASO audit, dry-run, customer reviews, TestFlight, in-app purchases, pricing, availability, age-rating, submit/release, and signing health. Anything else in the App Store Connect API is still reachable through the
raw_requestescape hatch — just ask.🛠️ Build & ship is different — it's an optional, local add-on. Almost every tool here talks to Apple's API and works from any machine. The three build tools (
bump_build_number,archive_app,upload_build) instead run local Xcode commands, so they only work on a Mac with Xcode installed and an Xcode project on disk. If those aren't present the tools tell you exactly what to install; everything else keeps working regardless.
list_apps → get the app idlist_app_store_versions (filter PREPARE_FOR_SUBMISSION) → the editable version idlist_app_store_version_localizations → the locale's localization idupdate_app_store_version_localization with keywords and/or descriptionIn practice you just ask: "set the keywords for MyApp to a, b, c" and the model chains these for you.
list_app_infos → list_app_info_localizations → update_app_info_localization.
list_app_store_version_localizations → list_screenshot_sets
(or create_screenshot_set) → upload_screenshot with an absolute image path.
Just ask: "Audit all my apps for ASO gaps." One audit_apps call returns a
ranked report — which apps are missing keywords, subtitles, descriptions, or
under-using the 100-char keyword field — plus an account-wide summary. Add
checkScreenshots: true to also flag apps with no screenshots.
// returns { summary: { appsAudited, appsWithIssues, issuesByType, … }, findings: [ … ] }
{ "name": "audit_apps", "arguments": { "checkScreenshots": false } }
Every update_* tool takes dryRun: true. You get the exact diff and length
checks, and nothing is written — so you (or the model) can confirm first:
{ "name": "update_app_store_version_localization",
"arguments": { "localizationId": "…", "keywords": "todo,tasks,planner", "dryRun": true } }
// → { dryRun:true, changes:[{ field:"keywords", from:"…", to:"…", newLength:18, limit:100, exceedsLimit:false }], warnings:[] }
Use raw_request, e.g. read customer reviews:
GET /apps/{id}/customerReviews. The whole API surface is reachable this way.
Metadata only saves on a version in an editable state (e.g.
PREPARE_FOR_SUBMISSION) — except promotional text, which can change live.
Edits go to the draft; they go public only after you submit and Apple approves.
.p8 or credentials. This repo's .gitignore and npm files whitelist are set up to prevent that.More: docs/SECURITY.md.
git clone https://github.com/fil-technology/appstore-api-mcp.git
cd appstore-api-mcp
npm install
cp .env.example .env # fill in your credentials
node src/index.js # runs the server on stdio (Ctrl-C to stop)
The server is plain ES modules, no build step. Source:
src/index.js — tool definitions + MCP wiringsrc/client.js — JWT (ES256) auth, request/paging helpers, asset uploadSome App Store essentials live outside App Store Connect and can't be set through this API — a hosted privacy policy or terms of use page, a contact form, or a way to collect user feedback.
AppMate generates hosted pages for exactly those (and has its own MCP server), so it pairs naturally with this one when you're filling out a listing.
Built by Sviatoslav (Sviat) Fil — filsv.com — indie iOS apps & developer tools.
MIT © Sviatoslav Fil — see LICENSE.
MCP server for Apple App Store Connect — edit listings (keywords, descriptions, titles, screenshots), track analytics (downloads, proceeds, subscriptions, retention), run a fleet-wide ASO audit, previ
{ "mcpServers": { "appstore-api-mcp": { "command": "npx", "args": [ "-y", "[email protected]" ] } }}Visão geral
MCP server for Apple App Store Connect — edit listings (keywords, descriptions, titles, screenshots), track analytics (downloads, proceeds, subscriptions, retention), run a fleet-wide ASO audit, preview changes with dry-run, and reach the full API. Works
Ferramentas
Ainda não há lista de ferramentas. Servidores remotos recebem uma após a primeira verificação de saúde.
Perguntas frequentes
Copie nesta página a configuração stdio para Cursor, Claude Desktop, Claude Code ou VS Code. Ela usa a instalação do publicador, não um proxy do PlugMCP.
Este registro não lista nenhuma variável de ambiente obrigatória.
Appstore Api roda localmente via stdio.
Última verificação: 2026-10-28. MCP server for Apple App Store Connect — edit listings (keywords, descriptions, titles, screenshots), track analytics (downloads, proceeds, subscriptions, retention), run a fleet-wide ASO audit, preview changes with dry-run, and reach the full API. Works