I do not think that an identification made only by comparison with NIST is enough. After you have a clue of the compounds present in or sample (by comparison with NIST), you should always confirm your results by injecting standards in the same analytical conditions (same column, same oven temperatures).

Concerning the series of n-alkanes you can run it separately.

Well, it is VERY important to use RI information for compound confirmation. You should make use of the two independent measurements, mass-spectra and retention (index) information, then you need a positive identification by spectra-similarity AND a similar retention behavior for an unequivocal identification. With only mass-spectra you might mix up structural similar compounds, relying only on RI makes even less sense.

However this requires a database containing RI information obtained under more or less identical conditions as yours, I am not sure if NIST can help you with this. As far as I know their compound library does not contain RI!? By the way, the GMD does, for different columns and RI methods. (http://gmd.mpimp-golm.mpg.de/download/)

As Clara said, measuring the compound to be identified as pure authentic standard on your system with your settings is usually more reliable than any database entries.

Regarding to your questions:

1.) Yes RI calculation works very fine even in complex matrices and is routinely done in many labs in the field of metabolomics.

2.) If you add your alkanes directly to your sample (usually during derivatization steps) helps to account even for tiny run-to-run variations. This is important if you want to measure multiple samples. Alkanes shouldn't be present in biological samples, so there is no problem, if you do not want to detect contaminations from a petrol company leaking into the eco-system.