Super Trouper

io.github.crissyfield由 GitHub 账号 crissyfield 发布v0.5.1更新于 Oct 24, 2026

MCP server for the Frida reverse engineering toolkit.

已验证STDIO仅桌面开发者工具安全与监控

概览

AI 生成的概览

面向 Frida 逆向工程工具包的 MCP 服务器，让助手检查设备、进程与内存并运行插桩脚本。

可以做什么
  • 列出并连接本地、USB 与远程 Frida 设备，启动、恢复或结束进程
  • 列出、查找并检查设备上的应用、进程、模块与线程
  • 附加到进程，创建、加载、卸载插桩脚本并收发消息
  • 读写小段内存，并在已附加进程中执行 JavaScript
  • 搜索、浏览并获取 Frida CodeShare 社区脚本
  • 管理 Frida Pattern 项目、向共享工作区安装 npm 包，并查询 Frida 版本与 API 文档
适用场景
适合逆向工程、动态分析与调试场景：需要助手在本地、USB 连接或远程目标上操作 Frida，获取社区插桩脚本，或查看运行中的进程状态。
运行要求
以本地 stdio 进程运行；可通过 npm 包 @crissyfield/super-trouper（npx）、Homebrew cask、Docker 镜像或发布版二进制安装。提供 macOS 与 Linux 预编译版本；从源码构建需要 Frida Core DevKit。未声明账号、API 密钥或环境变量。访问 CodeShare 与安装 npm 包需要网络。
安装前请注意
它会附加并修改运行中的进程：memory_write 会改写目标内存，device_kill 会终止进程，脚本会在目标内执行任意代码。它还能连接远程 Frida 服务器，并获取第三方 CodeShare 脚本与 npm 包，这些内容会在目标中执行。仅用于你有权分析的系统和应用。

安装

在 SourceWeft 中

  1. 打开 控制台中的 Super Trouper，将其添加到工作区。
  2. 为需要使用其工具的对话启用该服务。

Desktop only，通过 STDIO。 STDIO 服务会启动本地进程，因此需要 SourceWeft 桌面宿主。

其他 MCP 客户端

参照 仓库 中的启动说明。

README

[super-trouper logo]
[Build status] [npm version] [Go Reference] [MCP Rush score]

Super Trouper

An MCP server for the Frida reverse engineering toolkit.

Super Trouper runs an MCP server over STDIO that exposes Frida capabilities as tools for coding agents. It can connect to local, USB, and remote devices; inspect applications and processes; manage process sessions; and load, evaluate, and exchange messages with JavaScript instrumentation scripts. It can also search Frida CodeShare to discover community scripts and fetch them for use with the scripting tools. Device connections, sessions, and scripts are managed through the tools themselves and referenced by opaque handles.

[super-trouper]

Features

  • Single Go binary — no Python, Node, or Frida CLI tooling required on your host
  • Powered by Frida 17.x — statically linked against Frida Core DevKit v17.23.3
  • Full device support — list and connect to local, USB, and remote devices, attach to apps and processes
  • TypeScript and JavaScript instrumentation — create and load Frida scripts, evaluate directly in a target
  • Frida CodeShare integration — search, browse, and fetch the CodeShare catalog of community scripts
  • Runs anywhere — prebuilt binaries for macOS and Linux, install via Docker, Homebrew, or npm

Install

Homebrew

Install the macOS release through the project’s Homebrew tap:

sh
brew install --cask crissyfield/tap/super-trouper

npm

Install the launcher package through npm. It pulls in the platform-specific binary via optional dependencies:

sh
npm install -g @crissyfield/super-trouper

Release Binary

Download a binary for your platform from the Releases page.

From Source

Install a Frida Core DevKit matching your build host first. Then build and install Super Trouper with the following commands:

sh
# Set flags if the Frida devkit is not in a standard locationexport CGO_CFLAGS="-I/path/to/frida-core-devkit/include"export CGO_LDFLAGS="-L/path/to/frida-core-devkit/lib"
# Build and installexport CGO_ENABLED=1go install github.com/crissyfield/super-trouper@latest

Usage

Use the MCP server in your coding agent. Claude Code is the example below, but configuration is similar in Codex, OpenCode, Pi, Crush, and others.

Using the Binary

If the Super Trouper binary is not in your PATH, specify the full path in your MCP server configuration.

json
{  "mcpServers": {    "super-trouper": {      "command": "super-trouper"    }  }}

Using npx

npm users can run Super Trouper directly through npx without a global installation:

json
{  "mcpServers": {    "super-trouper": {      "command": "npx",      "args": ["-y", "@crissyfield/super-trouper"]    }  }}

Using Docker

Docker is a convenient way to run Super Trouper without installing it on your host.

json
{  "mcpServers": {    "super-trouper": {      "command": "docker",      "args": ["run", "-i", "--rm", "ghcr.io/crissyfield/super-trouper"]    }  }}

MCP Tools

Super Trouper exposes the following MCP tools:

ToolDescription
app_listLists the applications installed on a Frida device.
app_findFinds an application on a Frida device by identifier or name.
app_frontmostReturns the frontmost application on a Frida device.
codeshare_searchSearches Frida CodeShare for projects matching the given query.
codeshare_popularLists the most popular projects on Frida CodeShare.
codeshare_projectReturns a project from Frida CodeShare, including its JavaScript source, which can be run with the other scripting tools.
device_listLists the Frida devices detected by the host.
device_connectConnects to a Frida device by ID reported by device_list, by a remote Frida server address, or by type. Returns a device handle used by the other device tools; reconnecting to the same device returns the existing handle.
device_disconnectDisconnects from a Frida device. Fails while the device still has attached sessions.
device_paramsReturns the parameters of a Frida device.
device_spawnSpawns a process on a Frida device in a suspended state. Use device_resume to start it.
device_resumeResumes a suspended process on a Frida device.
device_killKills a process on a Frida device.
memory_readReads up to 4096 bytes of memory in an attached process and returns it as hex, base64, UTF-8, or a NUL-terminated C string. Fails if any byte of the range cannot be read.
memory_writeWrites up to 4096 bytes of memory in an attached process as hex (default) or base64, selected with the encoding parameter. Fails if memory cannot be written; patching code is out of scope.
module_listLists the modules loaded in an attached process. Results can be matched by name or path and paginated.
package_installInstalls an npm package in the shared Frida workspace for import from script_create source.
package_listLists direct npm packages installed in Super Trouper's shared Frida workspace.
pattern_createCreates a managed Frida Pattern project from .hexpat or .pat source files.
pattern_readReturns a source file from a managed Frida Pattern project.
pattern_listLists managed Frida Pattern projects.
pattern_deleteDeletes a managed Frida Pattern project.
pattern_describeCompiles a Pattern project for an attached target and returns diagnostics, types, and inputs.
pattern_decodeReads and decodes a bounded target-memory snapshot with a managed Frida Pattern project.
process_listLists the processes running on a Frida device.
script_createCreates a Frida script with TypeScript source in an attached session. The script is created unloaded; use script_load to load it.
script_loadLoads a Frida script into its target process.
script_unloadUnloads a Frida script from its target process.
script_closeUnloads a Frida script if loaded and releases its resources.
script_postPosts a JSON message to a Frida script.
script_messagesReturns messages sent by a Frida script. Messages are buffered (up to 256) and should be drained regularly.
script_bridge_listReturns the language bridges that can be exposed as globals in scripts created with script_create.
session_attachAttaches to a process on a Frida device by PID or name. Returns a session handle used by the session tools.
session_detachDetaches from a process, closing all of its scripts and its session.
session_evalEvaluates a JavaScript statement in an attached process and returns its JSON result. The first call sets up a persistent evaluator in the session.
frida_versionReturns the version of the linked Frida Core library.
frida_documentationReturns documentation links for the Frida JavaScript API.
thread_listLists the threads running in an attached process with their ID, state, and program counter.

来源：README.md，提交 d6dc245

工具

0
SourceWeft 曾在隔离沙箱中以无凭据方式启动该软件包一次，以列出其工具，但未能完成：服务器未能启动。

版本历史

4
  1. v0.5.1最新Oct 21, 2026
  2. v0.5.0Oct 15, 2026
  3. v0.4.0Sep 27, 2026
  4. v0.3.0Sep 18, 2026

评分

已满足适用检查项的 50%

该评分衡量此收录条目的完整程度，依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价，也不表示服务器安全或当前可用：可连通性是带日期的检查结果，评分也不包含安装量或用户评价。

  • 近 30 天内可连通25 分

    请保持端点能够响应 MCP 握手；SourceWeft 每 30 天探测一次。

  • 已声明认证方式10 分

    请在注册中心条目中声明凭据请求头，或保持端点能够响应探测，以便判定其认证方式。

  • 已列出工具15 分

    请让服务器在无凭据时响应 tools/list，或发布可在沙箱中无凭据启动的软件包。

  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分