XcodeBuildMCP

Why we included this project

Teams whose AI coding agents work on iOS or macOS projects often hit the gap between reading code and actually building and running it. XcodeBuildMCP closes that gap by exposing the Xcode build system to agents as structured tools: a model can compile for the simulator, prepare and run portable test products, and pull log output without you hand-typing long xcodebuild invocations or pasting opaque error dumps back into the chat. It ships in two forms, an MCP server that slots into Cursor, Claude Code, or Codex via small config snippets, and a plain CLI for terminal use, so you can keep the workflow you already live in. Optional agent skills prime a model on how to drive the tools, and a stateful daemon holds build and debug state across operations, which makes results noticeably less flaky than pointing an agent at raw command output. For teams that want agents to close the loop on their own Xcode builds and tests, it's a focused, pragmatic bridge rather than a general-purpose framework.