Version 6.0, March 2013
Please note that all references in this document to 'Windows' are to Windows 7 and earlier. I don't use Windows any more, and I have no idea whether this utility can still be built for Windows. Although it is very old, it still builds and runs fine on Linux./
KCrypt is a straightforward file encrypter and obfuscator, intended to provide strong security for sensitive files. It encrypts files using a key (password), making them incomprehensible until they are decrypted, using the same key. KCrypt can be used to secure any type of file on fixed or removeable storage devices. KCrypt also has a number of other privacy-related functions, as described below. KCrypt is based on the IDEA cipher, which is open, very strong, and no longer subject to patent restrictions in most countries.
KCrypt is primarily a command-line utility; versions are available for Linux, the native Windows command prompt, and Cygwin. However, there is also a version with a rudimentary graphical user interface, designed to be integrated with Windows Explorer -- more on this later. One of the main goals of this new release is cross-platform compatibility. Not only does the program operate the same way on all the platforms on which it can be built, but the files encrypted on one platform can be decrypted on another (see Platform compatibility notes below).
I originally wrote KCrypt to protect examination papers that I was preparing on my PC from the prying eyes of students, and for other confidential documents like letters of reference. Before I wrote this program, I would have kept these documents on floppy disks locked in my desk, which was not very convenient. These days, of course, desktop computers don't have floppy disk drives, and removing a hard disk at the end of each working day is even less convenient. KCrypt works best when encrypted files are kept together in a small number of directories, particularly when the same key is used for all files. This allows bunches of files to be encrypted and decrypted together in a batch.
KCrypt processes files in place; that is, the encryption process replaces the plaintext file with an encrypted version, removing all trace of the original file, while the decryption process has the opposite effect. In encrypting, some trouble is taken to ensure that no part of a plaintext file is left on disk or in the computer's memory after encryption. Thus you can encrypt a file in reasonable confidence that the unecrypted version is gone forever (until you decrypt it). However, a number of situations can subvert this security measure; these will be described in more detail below.
KCrypt can optionally rename the file to a random string of digits, so that the original name is not visible. In this case, it provides a facility to list the original names of encrypted files, provided you have appropriate password. KCrypt can also decrypt files to standard output, so they can be piped into another program.
KCrypt can process large batches of files in one invocation, and knows to avoid encrypting files that are already encrypted, or decrypting files that are not encrypted. This means that whole directories of files can be maintained in an encrypted state; it is easy to decrypt one or more files to work on them, then ensure that all files are encrypted by running KCrypt on the whole directory. It will even process an entire directory tree, although this kind of recursion has explicitly to be enabled on the command line.
KCrypt is a simple program with no external dependencies. It should be possible to build it on any platform that has a C compiler. There are no special installation requirements beyond copying the executable file to a convenient place; it does not need administrative privileges to run, and can even be installed without them if necessary. The overall size of the program is only about 100 kB and it uses negligible RAM at run-time. It's small enough that it should be possible to keep a version -- even several versions for different platforms -- on a memory stick along with the data it protects.
An encryption program, by its very nature, turns useful files into incomprehensible gibberish. It is to be hoped that decryption will restore them to a useful state; but, if either encryption or decryption fails, you will be left with useless files. As this would probably be a significant inconvenience, please test this program very carefully on non-critical data before trusting it on important files. I accept no responsibility whatever for undesired consequences of using this software. Although KCrypt has been available as a Windows program for over twenty years, and as a separate Linux program for over ten years, this new, fully cross-platform, version was essentially written from scratch. Please exercise the same caution as for any other relatively new software.
KCrypt was previously available as a shell extension (DLL) for the Windows desktop, and as a command-line program for Linux. These versions did not produce compatible files, as they were built using very different compilers. This version is different in a number of ways.
- The Windows GUI version is no longer a shell extension. It is a self-contained application, designed to be integrated with the Explorer 'Send to' menu (instructions below). The problem with shell extensions is that they can really only be created using Microsoft tools, and one of the main goals of this new versions is cross-platform compatibility. Despite being less tightly integrated with the desktop, the new version operates in much the same way as the old one did.
- This version of KCrypt does not advertise its existence
in files that it creates. A simple application of
stringswill not reveal the software used to encrypt the file, or even that it is encrypted. - While version 6.0 uses the same encryption algorithm as previous versions by default, it also offers a much faster (although weaker) algorithm for situations where speed is more important than security.
- Version 6.0 can be built for Cygwin under Windows, as well as Windows GUI and Windows console applications. The Cygwin version is more-or-less identical to the Linux version in operation.
- This version does not store the file modification time in the encrypted file, and restore it when the file is decrypted. In practice, doing this interfered with backup arrangements that relied on file timestamp.
Here are some examples of the use of KCrypt at the command prompt. For full details, please see the man page. In general, you can specify only a file or list of files, and the program will prompt for everything else; but the use of specific command-line switches is likely to be more convenient.
kcrypt6 -e docs/exams/december2013.doc
Encrypt the specified file in place (that is, overwriting the plaintext file with the encrypted version). Prompt for password.
kcrypt6 --recurse docs/exams
Prompt for what operation to carry out on the directory docs/exams,
then carry out that operation on the first 20 files found,
asking for passwords, confirmation, etc.,
as appropriate. The 20-file limit is a safety measure that applies to
recursive directory searches, and can be changed using
the --max-files switch.
kcrypt6 --max-files 100 --recurse -e docs/exams
Encrypt in place all the files in docs/exams, up to
a limit of 100 files. Prompt for
the password. Any files that are already encrypted (by KCrypt)
are skipped.
kcrypt6 -d *
Decrypt in place all the files in the current directory, and restore the original names if the filenames were mangled by a previous encrypt operation. Prompt for the password. Any files that are not encrypted (by KCrypt) are skipped.
kcrypt6 --xor -m private/letter.doc
Encrypt the specific file letter.doc and mangle its name, that
is, give it a random name in the same directory. Prompt for the
password.
kcrypt6 -l private/*
List the encryption status of all files in the private
directory. If a valid password is given when prompted, then
KCrypt displays the original filename of the file -- if
its name has been mangled -- and some additional, internal
information. If you don't care about the details, but just want
to know which files are encrypted, you can do:
kcrypt6 -l private/* --password ""
This is one of the few occasions where it is safe to specify the (null) password on the command line (for a discussion of this point, see below).
kcrypt6 -d docs/ --max-files 100
Decrypt and restore the original names of all encrypted files in
the docs directory, up to a limit of 100 files.
kcrypt6 -w *.txt --yes
Wipe (zero out and then delete) all files whose names end
in .txt, and
assume 'yes' is the answer to all prompts.
kcrypt6 --cat creditcard.txt
Decrypts the specified file and prints the result to standard output. No temporary files are created during the decryption process.
There is no specific installer for any platform -- KCrypt
is a single executable file called kcrypt6 (or
kcrypt6.exe on Windows/Cygwin). You can install that
file any place where it easy to find, and is on the search path
used by the command shell. On a Unix system,
/usr/bin is a good choice; on Windows nothing is as
obvious, but C:\windows will probably serve.
The Windows GUI version is called kcrypt6_gui.exe.
Note that this is not a full GUI -- it is intended to be integrated
into Windows explorer (see below).
KCrypt can be used with Windows Explorer, by adding it to the 'Send to' menu, which is accessible by right-clicking on a file or folder. The contents of the Send to menu are, among other things, the shortcuts found in a specific directory. On Windows 7 this directory is
C:\Users\[name]\AppData\Roaming\Microsoft\Windows\SendTo
On earlier versions it was
C:\Users\[name]\SendTo
To set up KCrypt, find the SendTo directory,
right-click and choose 'New Shortcut'. For the shortcut target,
select the location of kcrypt6_gui.exe. For the
name, choose whatever you want to appear on the menu. Thereafter,
you should be able to select files from the Explorer,
and use 'Send to' to send them to KCrypt.
Note that in GUI mode recursion is disabled, and you'll have to select specific files, or sets of files, to process. You can't select a directory. This is because the risk of disaster is just too high, given that KCrypt is not very diligent about asking for confirmations.
The GUI operation of KCrypt is intentionally very similar to the command-line, when the program is invoked with no switches. That is, the program will ask what operation you want to perform on the selected files, prompt for passwords, etc., just as at the command prompt.
KCrypt is written in ANSI-standard C, and intended to be compiled
with GNU gcc version 3 or later. The choice of compiler
is significant, because issues such as data type sizes and structure
alignment are very significant to the encryption process. There's no guarantee
that a version built by one version of GCC will produce encrypted files
that can be read by a version built by a different version. However,
there seems to be reasonable compatibility between GNU gcc
versions 3 and 4, on a range of different platforms.
Building the program should, in principle, be as simple as
unpacking the source bundle and running
make; you might need to define the environment
variable CC to point to gcc if your system does not
have a default for this. If you build with MinGW on Windows,
directives in the Makefile should cause both the command-line and
GUI versions to be built.
The standard algorthim provided by KCrypt is Lai and Massey's International Data Encryption Algorithm (IDEA). Although nearly thirty years old, this algorithm has proven very resilient in the field. There are theoretically stronger algorithms, but cryptographic strength is not really the limiting factor with a program of this sort (see discussion below).
The IDEA block cipher is used in a cipher feedback mode (CFB), to reduce the tendency of repeated patterns in the input data to give repeated patterns in the cipher text.
KCrypt can also use a simple XOR-based cipher, also in CFB mode, for non-critical situations where speed is more important that cryptographic strength. Even the simple XOR cipher will probably defeat all but the most determined intruders.
On a reasonable desktop computer, expect encryption and decryption to take about one minute per gigabyte with the IDEA algorithm under Linux. The native Windows (console or GUI) version is about half that fast, and the Cygwin version half as fast again, on the same hardware. The relatively poor performance of Cygwin is probably attributable to the overheads created by the Posix compatibility layer.
On a Motorola Xoom Android table, running at the command prompt, the encryption speed is about one-tenth the speed of a desktop PC, that is, about two megabytes per second. Perhaps not all that impressive, but faster than most Java-based encryption apps for Android.
Broadly, the XOR algorithm is about twice as fast as IDEA.
With care in compiler settings, it is relatively straightforward to build versions of KCrypt that are file-compatible between Linux, native Windows, and Cygwin on X86 hardware. The program can be built for Linux on ARM systems, but no guarantees can be given of file compatibility between different hardware platforms, particularly when the CPUs have different endian-ness.
A problem arises when moving files between platforms that use different character encoding for filenames. Although encryption and decryption should work, mangled filenames may not be restored correctly, because the stored filename may not be representible in the new platform's character set. In fact, handling Unicode filenames is generally problematic on Windows -- see 'General limitations' below.
KCrypt is not very persistent in asking for confirmation of
user actions. It assumes that if you select 1,000 files and
click the 'Wipe' button, that's exactly what you mean to do.
On the command line you can specify the --yes switch
to remove even the minimal prompting that is normally done.
KCrypt can only encrypt files in place (that is, overwriting the original file with an encrypted version), or to a new file with a random name in the same directory. By design, it won't allow files to be encrypted or decrypted into new files in other locations.
As a consequence of the point above, KCrypt needs the user to have access rights to create new files in the directory containing the files to be encrypted or decrypted.
This version of KCrypt is not backwards-compatible with any older version. I hear from time to time that there are a few people still using the very first version of KCrypt that I wrote for Windows 95 back in 1998. The problem is that compilers change over the years, not to mention operating systems, and I am not able to create a version of KCrypt that will read these old files with modern tools. It's probably possible, but there are only so many hours in the day.
A potential problem arises when handling Unicode filenames on the native Windows (not Cygwin) platform. When KCrypt stores the original filename in a mangled file, it uses the exact name provided by the C runtime library, byte-for-byte. On Windows, filenames are stored in UTF-16 format, but they are delivered to the program by the command prompt in ANSI (8-bit) equivalents. Consequently, any file whose name cannot be rendered using the ANSI character set will not be restored correctly when its name is unmangled. Using the GUI version of KCrypt for Windows does not entirely solve the problem, because ANSI filename conversion also happens in the interface between Windows and the C run-time library. Cygwin and Linux should not have this problem, and Windows is only affected for filenames that cannot be represented using the ANSI character set.
If you take the security of your data seriously, I recommend that you read this section very carefully.
KCrypt is designed to be convenient to use, while remaining reasonably secure. The IDEA algorithm it uses by default is still thought to be effectively unbreakable, but there are many ways to attack a program of this sort besides trying to break the cryptography. The biggest weakness is that, in order to prevent the user from accidentally trying to decrypt a file with the wrong password -- which would be very destructive -- encrypted files generated by KCrypt contain both encrypted and plaintext versions of a marker string. This allows the password to be tested easily by the software, because only the correct password will decrypt the encrypted marker string to match the plain-text version. But, of course, this set-up allows many passwords to be tested very rapdily, particularly on a fast machine. Any encryption software that provides a way for a computer (rather than a human) to determine whether the password is correct or not is open to dictionary-based brute-force attacks. Avoidance of dictionary words or other 'guessable' passwords will go a long way towards avoiding this sort of problem. However, it is important to understand that strong crytography does not, by itself, lead to impenetrable security. There's no question that providing a method to test whether the decryption key is correct has a detrimental effect on security; but it also makes operation very much more convenient.
Any cryptographic scheme can be compromised by a knowledgeable person who has access to your computer. For example, he or she could install a piece of software that records every keystroke you type, and sends them somewhere over the Internet. A knowledgeable person could even replace the KCrypt program file with another that looked just the same, but did something completely different. It might, for example, implement an encryption scheme that looked good, but was weak enough to crack, or responded to keys other than the one you know about. KCrypt is open-source, so you can, if you want, build it yourself from source (for Linux you will probably have to).
In general, KCrypt encrypts and decrypts to temporary files, then deletes unnecessary files at the end of the operation. 'Unnecessary' files in this case might include, for example, the original plain-text file when encrypting. The program attempts to block Ctrl+C and other interrupts, and to warn the user that exiting the program is likely to lead to inadequate cleanup of files. This blocking works better on some platforms than on others; on all platforms there is likely to be some way for a user to stop the program in its tracks -- pulling the plug out, in the last resort. To make errant temporary files easier to control, KCrypt only ever creates them in the same directory as the files it is working on.
Using KCrypt with the --cat switch to
display the contents of an encrypted file does not create any
temporary files of its own. However, if you pipe the output
into another application, then temporary files might get created as
part of the pipe process. This is particularly likely on Windows,
because the operating system does not have a built-in pipe
mechanism, so it simulates one using temporary files.
Files created this way are particularly
troublesome, because we can't be sure where they are. Of course,
Windows will delete them eventually, but only if you don't
exit KCrypt ungracefully; and even then it won't wipe the
contents before deleting.
KCrypt goes to some lengths to ensure that passwords are not left lurking in memory, where an intruder with access to your computer might be able to run a memory scanner to locate them. However, particularly with the Windows GUI, it is very difficult to be really sure that the information entered by the user is not retained somewhere in RAM or swap.
Similar considerations apply to temporary files deleted by the program -- KCrypt attempts to wipe all temporary files clean, by zeroing out their contents from start to finish. However, the software has no control over the low-level operation of the kernel and disk drivers, and it can offer no guarantees that some recognizable data won't be left lurking on the disk somewhere. Some filesystems even make a point of holding on to older versions of files, for the convenience of users who might accidentally delete or modify them.
For convience, KCrypt will allow you to throw caution to the wind and specify an encryption or decryption password on the command line, rather than entering at the prompt. The security implications of such an action hardly bear thinking about, and it would be unwise to do this for any purposes other than testing.
KCrypt only provides English-language messages and prompts.
KCrypt attempts not to echo passwords as they are typed.
There are specific ways to do this on native Windows, Cygwin,
and Linux platforms. On all platforms this interference with
the terminal/console can cause problems if you're redirecting
the output from the program to another utility that alters
terminal settings
(for example, using kcrypt6 -- cat | more).
The Windows GUI version of KCrypt does not allow
the encryption algorithm to be changed (IDEA is always used),
and it requires files, rather than directories to be selected.
There are no limts, however, other than those imposed by the
Windows desktop, on the number of files that can be selected.
Kcrypt intentionally does not process hidden files, if they are found in a directory search. However, if hidden files are specifically selected, either on the command line or through Windows Explorer, they are processed. The logic here is that if the user specifically selects a file it is intentional, but the user probably wouldn't want files being encrypted and decrypted when they are not normally visible.
The text 'OK', 'Error', and 'Skipped' in messages is highlighted in
colour if the output is to a terminal, unless the
--no-color switch is given. The Windows console
version does not support this colour highlighting, which is done
using direct output of ANSI escape codes. The Cygwin terminal, and
most Linux terminals, support these escape codes. Some other terminals
might not, but full support for all possibilities would require
use of the terminfo database and all the overheads associated with that.
KCrypt is maintained by Kevin Boone, and is open-source under the terms of the GNU Public Licence, version 3.0. There is no warranty of any kind. The IDEA algorithm was developed by Dr Xuejia Lai and Prof. James L Massey. A description of the algorithm may be found in: Lai, X (1992) ``On the Design and Security of Block Ciphers'', ETH Series on Information Processing (ed. Massey, JL) Volume 1, Konstanz, Switzerland: Hartung-Gorre Verlag, ISBN 3-89191-573-X. The legal status of the algorithm is not entirely certain, although the owners of the patent have always stated that they were happy for it to be used free of charge for private purposes. The original patents expired between 2011 and 2012 in most countries, but I cannot promise that it is patent-free in all jurisdictions.