Derivatization is not necessary for GC-MS analysis.
When conducting GC-MS analysis, derivate or not is dependent on the targeted molecule itself.
Typically speaking, some high polarity and non-volatile compounds that are difficult to gasify or show low ionization efficiency (such as organic acids and sugars) can be derivated to make the detection more sensitive.
your methanolic extract may contain polar molecules of large molecular mass and sometimes thermolabile molecules . I suggest you make a derivatisation and since silylation is the method of derivatisation is the most used method,First of all, you have to remove methanol by removing it from your extract because it is a protic solvent and therefore it destroys the silylant agent.so you need a dry extract from methanol and water.
The methanolic extract may contain for some high polarity and non-volatile compounds that are difficult to gasify or show low ionization efficiency (such as organic acids and sugars) can be deactivated to make the detection more sensitive.
Which class of compound are you looking for? For full metabolome coverage, you should inject both directly and after derivatization...If you are analyzing a certain compound or molecule group you should work on their molecular structure before determining the analytical strategy. Since TMS is the major derivatization approach you may change the reactant if the molecules are not bearing -OH. Be aware, that you better apply oximation initially (after drying MeOH extract) to stabilize the carbonyl moities prior to silylation.
A lot of correct stuff has been written just a few adds:
1) if You inject the methanolic solution Your collumn will not live very long (if You run in routine i.e. inject hundreds of samples a week. - if it is only one sample it does not really matter.
2) derivatisation might be needed depending on what You want to analyse. The classical sylilation agents will not work except the methanol is really free of water (very improbable). TMSH would still work usually:
there are some chloroformylations that might also work....
but in the overall: blow down the methanol, dissolve again in a non protic solvent and use a derivatisation agent of Your choice will probably be the most sustainable solution.