App Store Connect MCP
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
App Store Connect MCP
A Model Context Protocol server that exposes the entire Apple App Store Connect API (1,200+ operations) via 4 lightweight MCP tools. Query apps, manage builds, handle submissions, read analytics, manage users, and more — all from your AI assistant.
Why only 4 tools? Registering 1,216 tools would consume ~140K tokens of LLM context, making the server unusable. Instead, the LLM uses
search_apis→get_tool_details→call_apito discover and execute any operation on demand — keeping context tiny while maintaining full API coverage.
Features
Complete API coverage — all 1,216 operations from the App Store Connect API v4.4
Context-friendly — only 4 MCP tools registered (not 1,216), protecting LLM context window
Tool discovery workflow —
search_apis→get_tool_details→call_apiFull request/response schemas —
get_tool_detailsreturns the resolved JSON:API body schema (attributes, relationships with target resource types), the response shape, and a ready-to-copy example body — agents never fly blindResponse truncation — large API responses truncated at 25KB to fit LLM context
Two transports —
stdiofor local use (ggcode, Claude Desktop, Cursor),HTTPfor remote deploymentJWT authentication — automatic ES256 token generation and caching
Docker ready — multi-stage Dockerfile with health checks (TypeScript and Rust images)
Rust port — feature-parallel implementation in
rust/sharing the same generated catalog: single 6.6MB binary, no runtime depsAuto-publish — push a git tag
v1.x.xto publish to npm (OIDC, no token needed)
Related MCP server: App Store Connect MCP Server
Quick Start
1. Get API Credentials
Click Generate API Key (or use an existing one)
Note down:
Issuer ID — shown at the top of the Keys page
Key ID — shown in the key list
Private Key (.p8 file) — download it (only available once)
The key needs at least App Manager or Admin role for full functionality.
2. Install
Option A: npm (recommended)
npm install -g @ggaiteam/appstore-connect-mcpOption B: One-line installer
curl -fsSL https://raw.githubusercontent.com/topcheer/appstore-connect-mcp/main/scripts/install.sh | bashOption C: Docker
docker pull ghcr.io/topcheer/appstore-connect-mcp:latest3. Configure Your MCP Client
ggcode (topcheer/ggcode) — Recommended
Add to ~/.ggcode/ggcode.yaml under mcp_servers:
mcp_servers:
- args:
- -y
- '@ggaiteam/appstore-connect-mcp@latest'
command: npx
env:
APP_STORE_CONNECT_ISSUER_ID: your-issuer-id-here
APP_STORE_CONNECT_KEY_ID: your-key-id-here
APP_STORE_CONNECT_P8_FILE: /path/to/AuthKey_XXXXXXXXXX.p8
name: appstore-connect
type: stdioRestart ggcode. The 4 MCP tools (search_apis, get_tool_details, call_api, list_categories) will be available immediately.
Example session in ggcode:
> 列出我的 App Store Connect 应用列表
# ggcode will:
# 1. call_api("apps_getCollection", {arguments: {limit: 50}})
# 2. For each app, call_api("appStoreVersions_getToManyRelated", ...)
# 3. Summarize the resultsClaude Desktop
Edit ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"appstore-connect": {
"command": "npx",
"args": ["-y", "@ggaiteam/appstore-connect-mcp"],
"env": {
"APP_STORE_CONNECT_ISSUER_ID": "your-issuer-id",
"APP_STORE_CONNECT_KEY_ID": "your-key-id",
"APP_STORE_CONNECT_P8_FILE": "/path/to/AuthKey_XXXXXXXXXX.p8"
}
}
}
}Cursor
Add to your Cursor MCP settings:
{
"mcpServers": {
"appstore-connect": {
"command": "npx",
"args": ["-y", "@ggaiteam/appstore-connect-mcp"],
"env": {
"APP_STORE_CONNECT_ISSUER_ID": "your-issuer-id",
"APP_STORE_CONNECT_KEY_ID": "your-key-id",
"APP_STORE_CONNECT_P8_FILE": "/path/to/AuthKey_XXXXXXXXXX.p8"
}
}
}
}Alternatively, inline the private key instead of a file path:
"APP_STORE_CONNECT_PRIVATE_KEY": "-----BEGIN PRIVATE KEY-----\nMIGTAg...\n-----END PRIVATE KEY-----"
Restart your client.
Usage
Once connected, ask your AI assistant:
"List all my apps on App Store Connect"
"Show me the latest build for bundle ID com.example.myapp"
"Get all pending app review submissions"
"Search for subscription-related API operations"
The AI uses search_apis → get_tool_details → call_api to discover and execute operations.
4 MCP Tools
Tool | Description |
| Search 1,216 API operations by keyword, category, or HTTP method |
| Full parameter schema + resolved request-body schema + response shape + copy-ready example for one operation |
| Execute any API operation by name + arguments |
| List all 192 API resource categories with operation counts |
How It Works
User: "List my apps"
↓
AI calls: search_apis("apps")
→ finds: apps_getCollection (GET /v1/apps)
↓
AI calls: get_tool_details("apps_getCollection")
→ gets: query params, response shape (collection of apps,
fields: name, bundleId, sku, ...)
↓
AI calls: call_api("apps_getCollection", {arguments: {limit: 10}})
→ returns: [{name: "MyApp", bundleId: "com.example.myapp"}, ...]For POST/PATCH operations the details payload additionally contains the exact
JSON:API body schema — data.type enum, all attributes with types/enums,
and relationships with their target resource types — plus an example body
the model can copy and fill in.
Common Operations
What | Operation |
List apps |
|
List builds |
|
App versions |
|
Beta testers |
|
Sales reports |
|
Remote Deployment (HTTP Mode)
Docker Compose
# 1. Create .env file
cp .env.example .env
# Edit .env with your credentials
# 2. Start
docker compose up -dThe server is available at http://localhost:3000/mcp.
Docker (manual)
docker run -d \
--name appstore-connect-mcp \
-p 3000:3000 \
-e APP_STORE_CONNECT_ISSUER_ID=your-issuer-id \
-e APP_STORE_CONNECT_KEY_ID=your-key-id \
-e APP_STORE_CONNECT_PRIVATE_KEY="-----BEGIN PRIVATE KEY-----\n...\n-----END PRIVATE KEY-----" \
ghcr.io/topcheer/appstore-connect-mcp:latestTwo images are published on every push and release:
Image | Runtime | Size |
| Node 20 (TypeScript) | ~200MB |
| debian-slim (Rust binary) | ~146MB |
Both expose the same MCP surface: POST /mcp, DELETE /mcp, GET /health,
port 3000, HTTP transport by default. The Rust image additionally refuses
GET /mcp (server-initiated SSE) with 405, which MCP clients treat as
optional.
Health Check
curl http://localhost:3000/health
# {"status":"ok","server":"appstore-connect-mcp","version":"1.0.5"}Connect Remote MCP to Claude Desktop
{
"mcpServers": {
"appstore-connect": {
"url": "http://your-server:3000/mcp"
}
}
}CLI Usage
# Stdio mode (default — for local MCP clients)
appstore-connect-mcp
# HTTP mode (for remote deployment)
appstore-connect-mcp --transport http --port 3000 --host 0.0.0.0
# With verbose logging
appstore-connect-mcp --verboseCLI Flags
Flag | Default | Description |
|
| Transport mode: |
|
| HTTP port (HTTP mode only) |
|
| HTTP bind host |
| env | App Store Connect issuer ID |
| env | API key ID |
| env | Path to .p8 private key file |
| off | Enable verbose logging |
Environment Variables
Variable | Description |
| Issuer ID from App Store Connect |
| Key ID |
| PEM private key content (inline) |
| Path to .p8 file (alternative to above) |
|
|
| HTTP port |
| HTTP host |
|
|
Architecture
┌──────────────────┐ ┌──────────────────┐
│ MCP Client │ │ MCP Server │
│ (Claude, etc.) │◄───►│ │
└──────────────────┘ │ ┌────────────┐ │ ┌─────────────────┐
│ │ Tool │ │ │ App Store │
stdio or HTTP/SSE │ │ Registry │──┼────►│ Connect API │
│ │ (1,216 │ │ │ api.appstore │
│ │ tools) │ │ │ connect.apple │
│ └────────────┘ │ │ .com │
│ │ │ └─────────────────┘
│ ┌────▼─────┐ │ ▲
│ │ JWT Auth │ │ │
│ │ (ES256) │────┼────────────┘
│ └──────────┘ │
└──────────────────┘Tool data is generated from Apple's OpenAPI spec (
scripts/generate-tools.py)Runtime loads
tools.json(1,216 ops) but only registers 4 MCP toolsJWT tokens are cached and auto-refreshed (20-minute TTL)
Responses are truncated at 25KB to fit LLM context windows
Two Implementations
TypeScript ( | Rust ( | |
Runtime | Node >= 18, npm package | Single static binary |
MCP layer |
| Protocol implemented directly (JSON-RPC 2.0) |
HTTP transport | SDK StreamableHTTP (GET SSE wired, unused) | axum, |
Catalog |
| Same file embedded at compile time ( |
Package |
|
|
Env vars / CLI flags | identical | identical |
Both share src/tools.json as the single source of truth — regenerate it
with npm run generate and each implementation picks it up (TS: copy at
build; Rust: embed at compile).
Regenerating Tools
When Apple updates their API:
# Download the latest OpenAPI spec
# https://developer.apple.com/app-store-connect/
# Generate updated tools.json
python3 scripts/generate-tools.py openapi.oas.json src/tools.json
# Rebuild
npm run buildDevelopment
# Clone
git clone https://github.com/topcheer/appstore-connect-mcp.git
cd appstore-connect-mcp
# Install
npm install
# Generate tools from OpenAPI spec
npm run generate
# Build
npm run build
# Run locally
APP_STORE_CONNECT_ISSUER_ID=... \
APP_STORE_CONNECT_KEY_ID=... \
APP_STORE_CONNECT_P8_FILE=... \
npm startProject Structure
├── scripts/
│ ├── generate-tools.py # OpenAPI → tools.json generator
│ └── install.sh # One-line installer
├── src/
│ ├── index.ts # Entry point + CLI
│ ├── server.ts # MCP server (4 tools: search/details/call/list)
│ ├── transport.ts # stdio + HTTP transports
│ ├── auth.ts # JWT (ES256) token generation
│ ├── client.ts # App Store Connect API client
│ ├── tools.ts # Tool schema builder + executor
│ ├── types.ts # TypeScript type definitions
│ └── tools.json # Generated tool definitions (1,216 ops)
├── Dockerfile # Multi-stage Docker build
├── docker-compose.yml # Remote deployment config
├── .github/workflows/
│ ├── ci.yml # Lint + build + test
│ ├── npm-publish.yml # Auto-publish to npm on release
│ └── docker.yml # Build + push Docker image
└── package.jsonnpm Publishing
Publishing is fully automated via GitHub Actions with OIDC trusted publishing — no NPM_TOKEN needed.
Release a new version
git tag v1.0.3
git push origin v1.0.3That's it. The workflow will:
Extract version from the tag
Check if version already exists on npm (skip if so)
Build the package
Publish to npm with
--provenance=false --access publicAuto-create a GitHub Release
First-time setup (already done)
Package must exist on npm (first publish is manual:
npm publish --access public)Configure trusted publishing on npm: link the package to
topcheer/appstore-connect-mcpNo NPM_TOKEN secret needed — OIDC handles authentication
API Coverage
This server covers all 1,216 operations across the App Store Connect API:
Method | Count |
GET | 768 |
POST | 168 |
PATCH | 153 |
DELETE | 127 |
Including: Apps, Builds, App Store Versions, Beta Testing, Subscriptions, In-App Purchases, User Management, Sales Reports, Analytics, App Clips, Game Center, and more (192 categories total).
Security
Private keys are read from environment variables or files — never logged
JWT tokens are short-lived (20 minutes max) and cached in memory only
No data is stored or persisted between requests
For remote deployment, use HTTPS/TLS termination at your reverse proxy
License
MIT — see LICENSE
Acknowledgments
Apple App Store Connect API — API and OpenAPI specification
Model Context Protocol — Protocol specification
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
MCP server for your apps' tools and custom tools, plus hosted AI agents and approval-gated workflows
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to interact with yo…
ASOMobile MCP server: App Store Optimization data for the iOS App Store and Google Play inside AI agents. 29 tools: keyword research (traffic, competition, suggestions), app keyword indexing, competitors and keyword gaps, top charts and chart history, reviews with rating summaries, Apple Search Ads insights, organic download estimates and Keyword Monitor rank tracking, plus six ASO skills (app audit, keyword research, competitor audit, localization check, review analysis, ASA planning). Remote streamable HTTP with OAuth 2.1 (DCR + PKCE); sign in with your ASOMobile account.
Model Context Protocol server for the Apideck Unified API. Connect any MCP-compatible agent framework to 100+ accounting systems, HRIS platforms, file storage providers, and more through one integration. More information https://www.apideck.com/mcp-server
Related MCP Servers
- AlicenseNot gradedqualityBmaintenanceMCP server for App Store Connect API with 208 tools across 25 workers. Manage apps, builds, TestFlight, in-app purchases, subscriptions, reviews, provisioning, screenshots, analytics, and more — directly from Claude, Cursor, VS Code, or any MCP client. Multi-account support, worker filtering, JWT auth.84MIT
- AlicenseAqualityBmaintenanceA Model Context Protocol (MCP) server for Apple's App Store Connect API. Manage your iOS, macOS, tvOS, and visionOS apps directly from Claude, Cursor, or any MCP-compatible client.52186 npm23MIT
- AlicenseAqualityDmaintenanceA Model Context Protocol (MCP) server that connects Cursor, Claude Desktop, and other MCP clients to the official App Store Connect API—so you can manage iOS/macOS apps, TestFlight, in-app subscriptions, and store metadata via chat or automated tool calls.69120 npm14MIT
- AlicenseAqualityBmaintenanceApp Store Connect MCP server: your AI agent checks, fixes, submits and releases your iOS app. 41 tools: metadata, screenshots, builds, TestFlight, in-app purchases, preflight, submit, release, reviews, sales. Setup finishes in chat: paste your Issuer ID. Submit, release and price changes ask first; every error says what to do next. A fastlane alternative for AI agents. 7-day trial, no card.41845 npm2MIT