The practical answer really depends on two things: do you know exactly what you're looking for and can your instrument do tandem MS?
If you know exactly which metabolites you're looking for and they're present in great enough abundance, it may be possible to resolve them from the the background, depending on the resolving power of the instrument and what else is in there (if you have an ICR instrument, it may not be a problem at all, but if it's just a single quadrupole, you may be out of luck).
If you know exactly which metabolites you're looking for (or know that they share some common structural feature that can be knocked off by CID or the like) and your instrument can do tandem MS, you're in a much better position. Tandem MS is often used to avoid needing to do a slow LC step as it's possible to identify two compounds that aren't resolved on a precursor ion scan by looking at their fragment ions: A messy primary ion scan can be dealt with if you know that your analyte has a [M-18] transition or whatever, by CID. Also, a given class of compounds often shares certain CID reactions, which can be picked up with a neutral loss scan.
If you have tandem, but aren't sure exactly what you want to find, it's a bit more work, but it just means you need to look at product ion scans of peaks that are ambiguous and try to work out what they are.