:zap: Worlds fastest steghide cracker, chewing through millions of passwords per second :zap:
Stegseek is a lightning fast steghide cracker that can be used to extract hidden data from files. It is built as a fork of the original steghide project and, as a result, it is thousands of times faster than other crackers and can run through the entirety of
rockyou.txt* in under 2 seconds.
Stegseek can also be used to extract steghide metadata without a password, which can be used to test whether a file contains steghide data.
*rockyou.txtis a well-known password list with over 14 million passwords.
Skip ahead to Performance for some raw numbers.
The following instructions walk you through the installation process. Alternatively, you can run Stegseek in a Docker container. Skip ahead to Docker for instructions.
On Ubuntu and other Debian-based systems, you can use the provided
.debpackage for installation:
sudo apt install ./stegseek_0.5-1.deb
On other systems you will have to build Stegseek yourself. See BUILD.md for more information.
Building Stegseek as a native Windows app is sadly not supported. Instead you should run Stegseek using WSL. The Ubuntu WSL distribution is recommended for optimal compatibility.
Once you have configured WSL, Stegseek can be installed using the above Linux instructions.
By default, WSL mounts the
/mnt/c/, which you can use to easily access your files.
The most important feature of stegseek is wordlist cracking:
stegseek [stegofile.jpg] [wordlist.txt]
This mode will simply try all passwords in the provided wordlist against the provided stegofile.
Stegseek can also be used to detect and extract any unencrypted (meta) data from a steghide image. This exploits the fact that the random number generator used in steghide only has 2^32 possible seeds, which can be bruteforced in a matter of minutes.
stegseek --seed [stegofile.jpg]
This command will tell you: * Whether this file actually contains steghide content. * How much hidden content the file contains. * How the content was encrypted .
If you're (very) lucky and the file was encoded without encryption, this mode will even recover the encoded file for you!
The below demo features a challenge from X-MAS CTF 2020. A flag was hidden using a secure random password, but without encryption enabled. Within a few minutes, Stegseek is able to recover the embedded file without needing to guess the correct password.
stegseek --helpto get the full list of available options: ``` StegSeek version 0.5
=== StegSeek Help === To crack a stegofile: stegseek [stegofile.jpg] [wordlist.txt]
Commands: --crack Crack a stego file using a wordlist. This is the default mode. --seed Crack a stego file by attempting all embedding patterns. This mode can be used to detect a file encoded by steghide. In case the file was encoded without encryption, this mode will even recover the embedded file. Positional arguments: --crack [stegofile.jpg] [wordlist.txt] [output.txt] --seed [stegofile.jpg] [output.txt]
Keyword arguments: -sf, --stegofile select stego file -wl, --wordlist select the wordlist file -xf, --extractfile select file name for extracted data -t, --threads set the number of threads. Defaults to the number of cores. -f, --force overwrite existing files -v, --verbose display detailed information -q, --quiet hide performance metrics (can improve performance) -s, --skipdefault don't add guesses to the wordlist (empty password, filename, ...)
Use "stegseek --help -v" to include steghide's help. ```
Stegseek includes nearly all of steghide's functionality, so it can also be used to embed or extract data as normal. The only catch is that commands must use the
steghide embed [...]becomes
stegseek --embed [...].
You can also run Stegseek as Docker container:
docker run --rm -it -v "$(pwd):/steg" rickdejager/stegseek [stegofile.jpg] [wordlist.txt]
This does require that the wordlist and stegofile are located in current working directory, as that folder is mounted to
/steginside of the container.
This is where Stegseek really shines. As promised, let's start with the "
rockyou.txtin just 2 seconds" claim.
I picked the last password in
rockyou.txtwithout control characters: "␣␣␣␣␣␣␣1" (7 spaces followed by '1').
> time stegseek 7spaces1.jpg rockyou.txt StegSeek version 0.5 Progress: 95.21% (133213740 bytes)
[i] --> Found passphrase: " 1" [i] Original filename: "secret.txt" [i] Extracting to "7spaces1.jpg.out"
real 0m1,644s user 0m12,847s sys 0m0,041s
And there it is, over 14 million passwords in less than 2 seconds :heart_eyes:.
To test the performance of of other tools, I created several stego files with different passwords, taken from
rockyou.txt. I ran each of the tools with their default settings, except Stegbrute where I increased threading for a fair comparison.
| password | Line | Stegseek v0.5 | Stegcracker 2.0.9 | Stegbrute v0.1.1 (-t 8) | |-------------|-------------|----------------|-------------------|-------------------------| | "cassandra" | 1 000 | 0.04s | 3.1s | 0.7s | | "kupal" | 10 000 | 0.04s | 14.4s | 7.1s | | "sagar" | 100 000 | 0.04s | 2m23.0s | 1m21.9s | | "budakid1" | 1 000 000 | 0.06s | [p] 23m50.0s | 13m45.7s | | "␣␣␣␣␣␣␣1" | 14 344 383 | 1.65s | [p] 5h41m52.5s | [p] 3h17m38.0s |
[p] = projected time based on previous results.
In the first four examples, half of the time is spent syncing metrics between threads and displaying them. In quiet mode, StegSeek's numbers are:
[0.022s, 0.022s, 0.022s, 0.022s, 1.45s]
To compare the speed of each tool, let's look at the last row of the table (otherwise Stegseek finishes before all threads have started).
At this scale Stegseek is over 12 000 times faster than Stegcracker and over 7000 times faster than Stegbrute.
* Uses way less memory (memory usage is no longer tied to Wordlist length) * Wordlist is loaded on the fly, so we can start cracking immediately. * Fixed high false positive rate on
[email protected]:~/testfiles# time stegseek 7spaces1.jpg rockyou.txt -q Stegseek version 0.5 [i] --> Found passphrase: " 1" [i] Extracting to "7spaces1.jpg.out" real 0m0.837s user 0m12.694s sys 0m0.052s
> steghide --embed -sf steg.jpg -ef secret.txt -cf clear.jpg -f -p $(shuf -n 1 realuniq.lst) embedding "secret.txt" in "clear.jpg"... done writing stego file "steg.jpg"... done > time stegseek steg.jpg realuniq.lst - StegSeek version 0.5 Progress: 97.83% (15354873129 bytes) [i] --> Found passphrase: "Hossuungsstrahl" [i] Original filename: "secret.txt" [i] Extracting to stdout Now that I can stream in the file on the fly, I can run the cracker with stupidly large wordlists. This one is 15 GB! real 3m25,540s user 23m13,614s sys 0m10,260s
* Added seed cracking to allow for passwordless data extraction * Overhauled parser to allow for positional arguments * cracks
rockyou.txtin 2 seconds
Fixed a bug where stegseek would fail to find a password on rare occasions.
* Made threading lock-free s.t. multiple threads are used more efficiently * Added
--crack, used to specify the location of the extracted file * Made the .deb package compatible with Debian/Kali * cracks
rockyou.txtin 4 seconds
* Give up on a passphrase once a single bit of the magic fails to decode * Removed costly BitString / EmbData allocations * Improved performance of the selector * cracks
rockyou.txtin 5 seconds.
Initial release, features:
* Only loads the stego file once, unlike conventional crackers * Workpool based threading * Attempts to crack the first 3 bytes first, before attempting full decryption * Added .deb package and docker container * cracks
rockyou.txtin 41 seconds.