A mass spectrometer by itself, will tell you the mass to charge ratio of ions in a sample. In most samples the molecules themselves are not ions naturally and need to be ionized. This can be done using a number of techniques, but a common one is electron impact ionization, where electrons are slammed into molecules causing them to fragment into charged ions. This smashes the molecules up into small charged chunks which can then be analyzed by the mass spec. These mass spectra are good fingerprints for the original molecule and can be mapped back to what molecule must have been smashed to produced them. However, if you try to analyze the mass spectrum of a mixture of molecules that are similar you will not be able to deconvolute the result into any unique answer, i.e. there will be many possible mixtures of molecules that could have resulted in the mass spectrum.

This limitation of mass spectrometry can be overcome by first separating out the mixture of molecules you want to analyze into different fractions before they are ionized and enter the mass spectrometer. One common way to do this is to use a gas chromatography column. This works by first heating the sample so that it turns into a gas, and then passing it through a long column that is packed with a material that will interact with the gas and slow some molecules more than others. Often it is packed with silica. In this way you get the technique called GC-MS (gas chromatography/mass spectrometry).

GC-MS is a fantastic technique for analyzing complex mixtures but it will have its limits if there are too many different types of molecules in the mixture. In blood there will be tiny molecules like O2 and CO2 and giant macro-molecules like DNA. There is no technique I know of that can analyze such a wide range of molecular masses in one go. Plus mass spectrometry doesn't directly give you the structure just the masses of the fragments after you blow up the molecules (or the total mass if a more gentle ionization technique is used as is sometimes done for macromolecules).

I write all this just to explain how there aren't any analytical instruments that you just inject a sample into and out pops all the chemicals and their concentrations. Blood is a amazing soup of different elements and molecules and a full analysis requires a large number of steps and different analytical techniques to get even close to fully characterizing.

You'd typically need reference samples if you're trying to get a typical lab to analyze compounds via mass spectrometry, which would make identification of unknown substances in blood hard for non-researchers.

Mass spec (at least as of a few years ago) require reference spectra to identify compounds. Without them, you can guess at structure but it's not accurate.

Exactly, IBM can make movies using individual atoms, surely we aren't too far from being able to look at molecular structures and deduce what compounds are in a sample of blood.