re-frida
MCP server that wraps the Frida dynamic instrumentation toolkit, allowing users to attach to processes, hook functions, enumerate modules and exports, and manage scripts through natural language.
README
re-frida
MCP server wrapping the Frida dynamic-instrumentation toolkit. Frida injects a JavaScript engine (V8) into a target process and lets the analyst hook arbitrary functions, walk the type graph, and call into the target at runtime.
The MCP layer adds:
- a stable session identifier — multiple scripts and hooks can be installed under one session, sharing state.
- a strict allowlist of binary-operation shapes — Frida exposes the full V8 JS API to scripts but the MCP wrappers only call the canonical, well-understood primitives (attach, spawn, enumerate, hook, RPC).
- soft-skip behaviour — when the
fridaPython module or the nativelibfridaare missing, every tool returnsWARNwith an install hint and the plugin keeps working.
Tools
| Tool | What it does |
|---|---|
check_frida |
Health check — return frida version, native lib presence, USB device list |
start_session |
Spawn a new process under Frida (target: Android / iOS / native PID / remote endpoint) |
attach_pid |
Attach to a running process by host PID |
script_load |
Compile + load a Frida script (JavaScript) into a session |
script_call |
Call a method on a loaded script's exports (RPC) |
enumerate_modules |
List modules loaded into the session's process |
enumerate_exports |
List exports of a single module |
hook_method |
Install an Interceptor hook on a named method |
rpc_export |
Register a Python-side callable as an RPC export the JS side can call |
end_session |
Tear down a session, unload scripts, detach |
Install
Frida is a heavy install (the Python module pulls in frida +
frida-tools; the underlying libfrida is a native shared
library shipped via PyPI wheels). To install standalone:
pip install -e ./servers/re-frida
On the target device (the phone or VM Frida is talking to),
the matching frida-server binary must be running. See
https://frida.re/docs/.
Run
re-frida # stdio transport (default for MCP)
python -m re_frida # equivalent
Deferred to a future run
The original Explore findings called for a Frida server; this
plugin-internal scaffolding lands it so the deep-dive agents can
pick it up. The Windows targets in the Live Fire stress test
still use re-winedbg as the primary dynamic tool; the Android
target future run is where re-frida becomes the workhorse.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。