What is the Xlat project?
An open-source collection of (mostly Perl) tools and modules for doing general translation tasks.
Where is the Xlat project?
The published results are on CPAN, linked under individual modules (I'll start posting a table of contents when it's warranted). The development work itself is hosted on SourceForge. I know, I know, all the cool kids use Github these days. I haven't used it yet; it was already exciting enough to move from CVS to SVN last year. Get off my lawn!
What about OmegaT? (Or your favorite open-source tool.)
OmegaT is a translation tool, and is very useful. The Xlat project may end up including a translation tool (if so, it's going to be based on Wx and will lie far, far in the future), but that's not what it's about right now. Right now, I'm working on the low-level modules and tools that allow us to write automation tools.
An example: a terminology checker. Given File::TTX, we can load a TTX, then check its terminology against a termbase. This is a useful tool! So useful it's expensive if you pay for it (of course, if you pay for it you don't get egregious errors (see errors in Xlat, egregious, multiplicity of)). But this sort of script is the reason Xlat is being written.
Another example: you have a bunch of text files, say, oh, 228 of them. You want to translate them in one big TTX, then split the translation back out into 228 cleaned text files. File::TTX lets you do that.
These examples already work (caveat: see errors in Xlat, egregious, multiplicity of). This is the sort of thing that is helping me in my daily work.