Or anyone ever done it and have the file? It'd be pretty neat. I know you can search from this website, but being able to just type heisig anywhere to get the kanji would be a lot quicker and nicer. (Yes it's a crutch, but I'm doing it so I can look up the kanji in the first place)

If you don't get what I mean, let me show you an example
If I type 'kita' and hit space bar, I set it to give me 'ｷﾀ━━━━━━(ﾟ∀ﾟ)━━━━━━ !!!!!'
I could also do heisig words like type in 'wound' and get 傷 after hitting space bar. (although for the reading I'd put をうんｄ to make it simple)

For how to add your own, it depends on the OS, but for vista I know you can just go to the ▼ button on the IME and go to the dictionary tool or add entry to get where you need to go. (I had to create my own .dic, it wouldn't let me add it on the original)

I tried that but couldn't make it work with Windows XP's IME. The problem is that when you're not typing in kana mode it doesn't try to convert what you type. I haven't tried it yet with SCIM and the other Linux IMEs but it would probably be easier with those.

I also tried to locate an EDICT dictionary client that would accept Heisig keywords (http://forum.koohii.com/showthread.php?tid=1042) as its input method in the same way but apparently there is none.

Right, but you can just type in kana, so you end up with をうんｄ which can turn into 傷. Same thing, except you can't see what you're typing in romaji, just kana, but hey, even the Japanese type in a mix, like おｋ so it doesn't hurt to get used to it I guess　(And it's not like we really need to see what we type for just one word, right?)

The only problem is heisig with more than one word as you can't use space bar, but you can just leave the space out or use _ or -

There is a way around it to see what your typing though, you can change it to Ｗｏｕｎｄ as 傷, and when you type Ｗｏｕｎｄ press F7 then press space bar, but this is a bit of a hassle (although still easier than actually looking it up elsewhere)

Also, it will work with multiple words, I can change Ｔｏｏ　ｍｕｃｈ to 余

Yeah that would work (though I suppose a collision of two keywords resulting in the same output could happen it's probably unlikely). You would have to write a script to generate the dictionary with all the keywords properly converted to pseudo-Japanese though.