Model Context Protocol Server

DevTools MCP Server

A Python MCP server that gives AI coding agents structured access to builds, simulators, emulators, tests, screenshots, logs, audits, and safe local automation.

DevTools MCP Server connects AI agents to local development tools, so agent workflows can build projects, inspect simulator and emulator state, trigger Xcode, run tests, inspect logs, capture screenshots, and run audits through a structured MCP interface.

The current tool surface covers iOS/Xcode, Android, and workflow automation, with planned expansion into web, backend, release, coverage, security, and runtime analysis tools.

What this demonstrates: practical agent tooling around real developer workflows, including host-machine automation, structured build errors, multi-project configuration, bounded local CI commands, and remote SSE access for container-to-host setups.

MCP
Python
Server
Tool registry for agents
Apple
iOS
Xcode Tools
Build, simulator, screenshots, logs
Android
Gradle
Mobile Tools
Build, devices, tests, lint
Workflow
CI
Local Automation
Audits and approved commands

Available Now

iOS

iOS And Xcode

Builds Apple-platform projects, lists simulator targets, captures simulator screenshots, returns bounded simulator logs, and can trigger Xcode run/stop through AppleScript.

Android

Android Automation

Runs Gradle builds, lists emulators and devices, manages app lifecycle, runs unit or instrumented tests, parses JUnit results, and exposes lint and Kotlin hygiene checks.

Workflow

Safe Workflow Tools

Checks changed files before commit, lists configured projects, runs only approved local CI commands, and manages the LLM Gateway during local development.

SSE

Container-To-Host Access

SSE mode lets container-based agents call host-machine tools like Xcode on the Mac without pretending those tools are available inside the container.

Quality Checks

Swift

Swift Hygiene

Checks for print, try!, fatalError, TODO, and FIXME before code reaches a commit.

Design

Design-System Checks

Flags hardcoded colors, raw RGB/hex values, and system font usage in changed files.

Meta

File Metadata Checks

Looks for unwanted generated headers and project noise that should not be committed.

Locale

Localization Support

Supports xcstrings read/write flows and localization audits for Apple-platform projects.

Kotlin

Kotlin Hygiene

Checks for println, !! null assertions, and TODO/FIXME comments in Android work.

Errors

Structured Results

Keeps build errors compact enough for an AI agent to understand and act on directly.

Technology Stack

MCP Server

Python Model Context Protocol SSE transport pytest

Developer Tooling

xcodebuild Xcode AppleScript iOS Simulator Gradle Android Emulator JUnit

Planned Tool Areas

Web

Web Apps

Planned tools for Vite, Next.js, React builds, previews, browser tests, deploy checks, Lighthouse, and bundle analysis.

Server

Server And Backend

Planned tools for REST or GraphQL tests, Docker, Docker Compose, bounded SSH deploy steps, and server health checks.

Android

Android Improvements

Android is available now, with logcat filtering and ProGuard/R8 release checks still planned.

Release

Release And Quality

Later ideas include coverage reports, security scans, rollback support, and crash or log analysis.

Repository Entry Points

The repository is organized around an MCP server entry point, platform drivers, audit checks, approved local automation, and setup documentation for local and remote agent use.

Review the MCP implementation, available iOS and Android tooling, quality checks, and setup path for connecting AI agents to local development workflows.