This repository contains scripts for working with Frida on Windows. The objective of Frida scripts is to inject into a third-party target process and modify behavior.
Scripts are provided for interacting with COM, WinRT and .NET APIs directly from javascript Frida scripts.
- Learn more about Frida, a dynamic code instrumentation toolkit.
- Install Frida
npm install -g frida frida-compile
- Review the Frida JavaScript API, which provides access to native functions and memory manipulation. Scripts here are based on this API.
The workflow is this:
- A target app is running on the machine
- A crafted javascript file (the script) compiled using
frida-compile
, merging in common resources. - The script is attached and injected into the target (e.g.
frida -p 1234 -l myscript.compiled.js
) - A custom set of instructions in the script manipulate memory in the target using Frida APIs, as well as higher level APIs like com.js and dotnet.js to call more complex APIs (e.g. opening a named pipe or streaming data to a log file)
- The target app and script both operate concurrently from the process space of the target
We attach to a running instance of notepad by looking in task manager for the PID
(say 1447
in this case):
frida -p 1447 -l myscript.compiled.js
Frida will then start and attach to the target:
____
/ _ | Frida 11.0.12 - A world-class dynamic instrumentation toolkit
| (_| |
> _ | Commands:
/_/ |_| help -> Displays the help system
. . . . object? -> Display information about 'object'
. . . . exit/quit -> Exit
. . . .
. . . . More info at http://www.frida.re/docs/home/
Attaching...
[Local::PID::1447]->
At this point as long as no errors are present in the output, the script is attached and ready to go. Scripts in this repository usualy print Begin
or Ready
to signal the script is actually loaded. If frida quits, there may be a parse error in the script.
- The Frida console has commands and features, but we don't need to use it for scripts here.
- You can also launch and attach to processes using other methods.
Fix-SHBrowseForFolder replaces the legacy folder dialog with the modern new dialog, enabling path entry.
(Left) Legacy SHBrowseForFolder dialog, (Right) modern IFileDialog dialog.
Group a specific window differently on the taskbar by assigning a unique identity:
Undo fullscreen when a modern app attempts to enter fullscreen mode. Hook WinRT API.
Verify that DotNetBridge is working properly by exercising calling .net APIs from the system and a locally compiled library.
Verify WinRT APIs are working.
Define a COM or WinRT interface:
var CLSID_FileOpenDialog = GUID.alloc("DC1C5A9C-E88A-4dde-A5A1-60F82A20AEF7");
var IFileDialog = new COM.Interface(COM.IUnknown, {
Show: [0, ['uint']],
SetOptions: [6, ['uint']],
GetResult: [17, ['pointer']],
}, "42f85136-db7e-439c-85f1-e4075d135fc8");
Each entry value is an ordinal in the interface vtable and a set of function argument types.
Create an object and work with the resulting COM interfaces:
COM.Initialize(COM.ApartmentType.STA);
var modalWindow = COM.CreateInstance(CLSID_FileOpenDialog, COM.ClassContext.InProc, IFileDialog);
modalWindow.SetOptions(FOS_PICKFOLDERS);
modalWindow.Show(browseinfo.hwndOwner);
var shellItem = new COM.Pointer(IShellItem);
COM.ThrowIfFailed(modalWindow.GetResult(shellItem.GetAddressOf()));
var pidl = Memory.alloc(Process.pointerSize);
COM.ThrowIfFailed(SHGetIDListFromObject(shellItem.Get(), pidl));
Call .NET APIs directly from javascript.
const CLR = require("../common/dotnet");
const System = new CLR.Namespace("System");
System.IO.File.WriteAllText(path, "log data");
Create a BROWSEINFOW struct at browseinfoPtr
:
var browseinfo = new Struct({
'hwndOwner':'int',
'pidlRoot':'pointer',
'pszDisplayName':'pointer',
'lpszTitle':'pointer',
'ulFlags':'uint',
'lpfn':'pointer',
'lParam':'long',
'iImage':'int',
}, browseinfoPtr);
Then read or write as object properties:
console.log("Flags: 0x" browseinfo.ulFlags.toString(16));