[boardstretcher]

Agreed.

I would add 'grep' in there as an absolute must have.

The sed, grep, awk holy trinity will allow you to do pretty much any type of text manipulation and searching that you can imagine.

[aethertap]

I have one more that I like. I call it 'rex' and it basically allows me to do a regex substitution on command line text (not stdin, but actual command line args). It mostly comes in handy when looping over files in a directory where I want to do some kind of processing based on information in the file name. For example, I can tag all of the episodes of a podcast with their track numbers (several podcasts I listen to don't do id3tags) with this command:

    for f in *.mp3; 
      do id3v2 -T `rex 's/^[^0-9]*([0-9]+).*\.mp3$/\1/` $f; 
    done

The script itself is really simple, but I can post the source if anybody's interested.