An App Store Connect tool for AI coding agents
asc-mcp
An open-source MCP server: ship an app to the App Store without leaving Claude Code or Cursor.
Where it runs
- npm
- live
Since April 2026
Run a preflight check on my app
release_preflightMetadata, character limits, screenshots and build status audited.
Give me a morning briefing
daily_briefingWho is in review, who was rejected, new low ratings, and what to do.
Generate release notes from my git history
release_notesCommits since the last tag, grouped for the What's New text.
Who it is for
iOS developers who work with an AI coding agent and do not want to open the App Store Connect portal for every release.
What it does
- 41 tools: metadata, screenshots, builds, TestFlight, in-app purchases, submission and release.
- A check before submission: character limits, screenshots, build status.
- A morning briefing across all your apps, including new low ratings.
Business model
Open source under MIT, with a free trial and a paid Pro plan.
Public figure
996 npm downloads in the 30 days to 11 September 2026.
What is hard to copy
Curated tools, not a raw API list
41 tools, each built around a release task, that already know App Store Connect's rules, such as character limits and what a submission needs.
No model cost on the server
Two tools use MCP sampling: the user's own model does the AI work, so the server pays for no model and resells nothing.
No fastlane, no Ruby
fastlane talks to the same App Store Connect API. So nearly everything you would script with fastlane is a tool here, and a Mac is only needed to build and sign the binary.
Why it is built this way
Setup without editing JSON
In Claude Desktop it installs with one click and the key is picked in a file dialog. Other clients are configured by one command that finds the key itself.
The key stays on your machine
The server runs locally and talks to Apple's API directly. Six tools are free, with no sign-up.
Have something like this in mind?
Tell me what it should do. On a 30 minute call we work out whether it is worth building and what the first version needs.