原型 · 预览版:收录官方注册中心里的全部服务器,每 30 分钟同步一次。评价和账户功能将在后续阶段推出。
PopNx MCP为使用 AI 的人准备的 MCP 服务器 · 用大白话讲清楚

Shuttle

作者:grarcht版本 4.0.0i★ –i在你的电脑上运行
注册名称io.github.grarcht/shuttle

安装

概览

作者提供的说明(英文原文)

Android framework and solution that prevents TransactionTooLargeException crashes by design.

PopNx MCP 的三大支柱

这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。

它能帮你做什么

我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。

  • 连接方式: 在你的电脑上运行
  • 传输方式: stdio
  • 分类: 设计、图片与音频

安装

选择你使用的 AI 应用,然后复制设置。

安装之前

  • 基础检查:1 项中通过 1 项
  • 在你的电脑上运行,拥有你的用户账户的权限(文件、网络、应用)。
  • 注册中心没有列出需要填写的密钥。

你需要

你用的是哪个应用?

在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。

打开 Claude Desktop → Settings → Developer → Edit Config,把下面的内容粘贴到 claude_desktop_config.json

{
  "mcpServers": {
    "shuttle": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "ghcr.io/grarcht/shuttle-mcp:4.0.0"
      ]
    }
  }
}

在终端中运行这条命令

claude mcp add shuttle -- docker run -i --rm ghcr.io/grarcht/shuttle-mcp:4.0.0

在 Cursor 中打开

或者自己复制设置

粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json

{
  "mcpServers": {
    "shuttle": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "ghcr.io/grarcht/shuttle-mcp:4.0.0"
      ]
    }
  }
}

浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。

在 VS Code 中打开

或者自己复制设置

粘贴到项目中的 .vscode/mcp.json

{
  "servers": {
    "shuttle": {
      "type": "stdio",
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "ghcr.io/grarcht/shuttle-mcp:4.0.0"
      ]
    }
  }
}

浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。

这个服务器在你的电脑上运行,而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。

你的选择通过 Cookie 保存在这台设备上,无需账户。

这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。

它能做什么

它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。

基础检查i

我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月6日

  • ✓ 通过

    作者与仓库一致

    注册名称已通过 GitHub 账户 grarcht 验证,该账户拥有此仓库

  • … 尚未检查

    仍在维护

    我们还没有从 GitHub 读取这个仓库

  • … 尚未检查

    有许可证

    我们还没有从 GitHub 读取这个仓库

  • – 不适用

    端点有响应

    它在你的电脑上运行,没有互联网端点

  • – 不适用

    使用 OAuth 登录

    在你的电脑上运行,改用本地设置

  • … 尚未检查

    工具描述中没有隐藏指令

    它在你的电脑上运行;我们只能读取互联网端点的工具列表

与“设计、图片与音频”中的其他服务器比较i

  • 星标数多于此分类中其他 2169 个服务器的 0%
  • 发布到注册中心的时间比此分类中其他 2169 个服务器的 20% 更新
  • 基础检查结果等于或优于此分类中其他 2169 个服务器的 100%

与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。

常见问题

Shuttle 安全吗?

它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。

它需要哪些密钥或代码?

注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。

它在我的电脑上运行还是在云端?

它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。

它还在维护吗?

GitHub 仓库最后更新于–,最新版本(4.0.0)于3个月前进入注册中心。

哪些应用可以使用它?

支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Docker。

有什么可以替代它?

同一分类中有:Figma Context MCP, Claude Real Video, Figma MCP Server

替代选择

同一分类中其他作者的服务器,按星标数排序。

如何看卡片:★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i

Figma Context MCP

作者:GLips

Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.

在你的电脑上运行★ 16,165MIT5个月前更新
安装

Claude Real Video

作者:HUANGCHIHHUNGLeo

Let any LLM watch a video locally — and search everything it has ever watched.

在你的电脑上运行★ 2,284MIT2个月前更新
安装

Figma MCP Server

作者:figma

The Figma MCP server brings Figma design context directly into your AI workflow.

在云端运行★ 2,15213个月前更新
安装

作者的 README

打开 README

按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。

Prevent TransactionTooLargeException crashes. For good. A modern Android framework for safely transporting large Serializable objects between components, without the crashes.


Get To It Quick

  • API Documentation
  • Why Shuttle?
  • How It Works
  • Requirements
  • Quick Start
  • Usage
  • Transport with Intents
  • Transport with the Navigation Component
  • Pick Up Cargo at the Destination
  • Cargo States (LCE Pattern)
  • Using @ShuttleCargo
  • Cleaning Up
  • Architecture
  • Context Diagram
  • Module Overview
  • Module Dependency Diagram
  • Demo Apps
  • Heads Up: Know the Tradeoffs
  • AI
  • Model Context Protocol (MCP)
  • Contributing
  • License

🚨 Why Shuttle?

🗞️ Featured in [Android Weekly #594](https://androidweekly.net/issues/issue-594) and [Android Weekly #455](https://androidweekly.net/issues/issue-455), validated by the Android community twice.

You've seen this before. Maybe last week:

android.os.TransactionTooLargeException: data parcel size X bytes

It didn't show up in dev. It didn't show up in QA. It showed up at 2am, in production, for real users. Your Play Store rating took the hit before anyone on the team even knew.

So you triaged it. Filed the ticket. Wrote the fix. Reviewed the PR. Ran QA again. Cut the hotfix. And then you added it to the code review checklist, hoping the next engineer would catch it before it happened again.

They won't. Not reliably. You can't review your way out of a structural problem.

Shuttle provides a modern, guarded way to pass large Serializable objects with Intent objects or save them in Bundle objects to avoid app crashes. The crash class is structurally prevented, not governed against.

Why keep spending more time and money on governance through code reviews? Why not embrace the problem by providing a solution for it?

Shuttle reduces the high level of governance needed to catch `TransactionTooLargeException` inducing code by:

  1. storing the Serializable and passing an identifier for the Serializable
  2. using a small-sized Bundle for binder transactions
  3. avoiding app crashes from TransactionTooLargeExceptions
  4. enabling retrieval of the stored Serializable at the destination

Shuttle also excels by:

  1. providing a solution with maven artifacts
  2. providing Solution Building Blocks (SBBs) for building on
  3. saving time by avoiding DB and table setup, especially when creating many tables for the content of different types of objects

When envisioning, designing, and creating the architecture, quality attributes and best practices were kept in mind. These attributes include usability, readability, recognizability, reusability, maintainability, and more.

Without ShuttleWith Shuttle
Large Serializable passed in Intent/BundleObject stored in a warehouse; only a small identifier is passed
Silent in dev, catastrophic in productionBinder transaction stays within safe size limits, everywhere
Time and money spent on crash investigation, fixes, QA, and hotfixesCrash class is structurally impossible
Requires constant code review governanceShip with confidence
Engineers manually manage object lifecyclesAutomatic or on-demand cargo cleanup built in

⚙️ How It Works

The Shuttle framework takes its name from cargo transportation in the freight industry. Moving and storage companies experience scenarios where large moving trucks cannot transport cargo the entire way to the destination (warehouses, houses, et cetera). These scenarios might occur from road restrictions, trucks being overweight from large cargo, and more. As a result, companies use small Shuttle vans to transport smaller cargo groups on multiple trips to deliver the entire shipment.

After the delivery is complete, employees remove the cargo remnants from the shuttle vans and trucks. This clean-up task is one of the last steps for the job.

The Shuttle framework takes its roots in these scenarios:

  • creating a smaller cargo bundle object to use in successfully delivering the data to the destination
  • shuttling the corresponding large cargo to a warehouse and storing it for pickup
  • linking the smaller cargo with the larger cargo by an identifier
  • providing a single source of truth (Shuttle interface) to use for transporting cargo
  • providing convenience functions to remove cargo (automatically or on-demand)

Shuttle applies this same logic to Android's binder transaction limit:

┌─────────────────────────────────────────────────────────────┐

│ Source Component │ │ 1. Large Serializable -> stored in Warehouse (Room/DB) │ │ 2. Small cargo ID -> passed in Intent/Bundle │ └──────────────────────────────┬──────────────────────────────┘ │ (tiny binder transaction) ┌──────────────────────────────▼──────────────────────────────┐ │ Destination Component │ │ 3. Cargo ID received -> retrieved from Warehouse │ │ 4. Large Serializable -> delivered via Kotlin Channel │ │ 5. Cleanup -> cargo removed from Warehouse automatically │ └─────────────────────────────────────────────────────────────┘


Requirements

RequirementMinimum
Android min SDK26
Kotlin2.2.10
AGP (Android Gradle Plugin)8.0+
KSPRequired only when using @ShuttleCargo
Java21

🚀 Quick Start

1. Add Dependencies

`settings.gradle.kts` — apply the Shuttle Gradle plugin so the @ShuttleCargo annotation processor is wired into your build:

@ShuttleCargo annotates the cargo class you want to transport between screens. At build time, the annotation processor generates the serialization code Shuttle needs to store and retrieve cargo, without any manual Serializable boilerplate.

pluginManagement {

plugins { id("com.grarcht.shuttle.cargo") version "4.0.0" } }

The Shuttle Cargo Gradle plugin configures KSP and registers the Shuttle compiler plugin automatically. Without it, @ShuttleCargo-annotated classes will not generate the required serialization glue code.

`build.gradle.kts`:

plugins {

id("com.grarcht.shuttle.cargo") }

dependencies { implementation(platform("com.grarcht.shuttle:shuttle-bom:4.0.0")) implementation("com.grarcht.shuttle:framework") implementation("com.grarcht.shuttle:framework-integrations-persistence") implementation("com.grarcht.shuttle:framework-integrations-extensions-room") implementation("com.grarcht.shuttle:framework-addons-navigation-component") // Optional

// Annotation-based API — use @ShuttleCargo to own the serialization contract implementation("com.grarcht.shuttle:framework-annotations") ksp("com.grarcht.shuttle:framework-annotations-processor") }

Version Catalog (`libs.versions.toml`):

[versions]

shuttle = "4.0.0"

[plugins] shuttle-cargo = { id = "com.grarcht.shuttle.cargo", version.ref = "shuttle" }

[libraries] shuttle-bom = { group = "com.grarcht.shuttle", name = "shuttle-bom", version.ref = "shuttle" } shuttle-framework = { group = "com.grarcht.shuttle", name = "framework" } shuttle-persistence = { group = "com.grarcht.shuttle", name = "framework-integrations-persistence" } shuttle-room = { group = "com.grarcht.shuttle", name = "framework-integrations-extensions-room" } shuttle-navigation = { group = "com.grarcht.shuttle", name = "framework-addons-navigation-component" } shuttle-annotations = { group = "com.grarcht.shuttle", name = "framework-annotations" } shuttle-annotations-processor = { group = "com.grarcht.shuttle", name = "framework-annotations-processor" }

2. Initialize Shuttle

Wire CargoShuttle once as a singleton. With Hilt:

@Provides @Singleton

fun provideShuttle(facade: ShuttleFacade, warehouse: ShuttleWarehouse): Shuttle = CargoShuttle(facade, warehouse)

@Provides @Singleton fun provideShuttleFacade( @ApplicationContext context: Context, warehouse: ShuttleWarehouse ): ShuttleFacade = ShuttleCargoFacade(context as Application, warehouse)

@Provides @Singleton fun provideShuttleWarehouse( dao: ShuttleDataAccessObject, factory: ShuttleDataModelFactory, @ApplicationContext context: Context, gateway: ShuttleFileSystemGateway ): ShuttleWarehouse = ShuttleRepository(dao, factory, context.filesDir.absolutePath, gateway)

The ShuttleDataAccessObject comes from ShuttleRoomDataDb.getInstance(ShuttleRoomDbConfig(context)).shuttleDataAccessObject. Inject Shuttle wherever you need to transport data.

3. Ship Your First Cargo
// Source: transport a large Serializable via Intent

shuttle.intentCargoWith(context, DestinationActivity::class.java) .transport(cargoId, myLargeSerializable) .cleanShuttleOnReturnTo(SourceFragment::class.java, DestinationActivity::class.java, cargoId) .deliver(context)

// Destination: pick up the cargo

lifecycleScope.launch { getShuttleChannel() .consumeAsFlow() .collectLatest { result -> when (result) { is ShuttlePickupCargoResult.Success<*> -> render(result.data as MyModel) is ShuttlePickupCargoResult.Error<*

只显示了开头部分;其余内容请在仓库中阅读

问 PopNx