Hey, I'm not sure you can gain access to the raw data by the api, however, I think file=100.2 contains the reads/contigs that passed quality filtering. There's probably also a file that contains the reads/contigs that didn't pass QC, so you could combine those if you really wanted them. You could always ask at the mg-rast mailing list..

File 050.2 - This is the unfiltered metagenome that was originally uploaded to MG-RAST
File 100.1 - preprocess.passed.fna
File 100.2 - preprocess.removed (low quality)
File 350.2 & 350.3 - These are the protein coding genes (amino acids and nucleotides)
File 440.1 - These are predicted rRNA sequences (I do not recommend using MG-RAST for sensitive rRNA annotation. It does not use the internal structure of the gene, which other programs appropriately use for classification)
File 550.1 - This file shows clustered sequences which are 90% identical, to reduce the number of sequences that need to be annotated. Many folks don’t even know that this happens within MG-RAST.
File 650.1 & 650.2 - These files are essentially the blat tabular output from comparing your sequence to the database.

see example: http://metagenomics.anl.gov/metagenomics.cgi?page=DownloadMetagenome&metagenome=4447943.3

http://api.metagenomics.anl.gov/1/download/mgm4447943.3

ref: http://adina-howe.readthedocs.io/en/latest/mgrast/index.html

It's Bash. I presume you could get Bash working on Windows with Cygwin or something but I haven't used Windows since XP so I can't really say. Alternatively, it probably doesn't take much effort to make a simple while read line loop with something that works on Windows by default like Python (?) or Java (?). If you plan to do lots of bioinformatics in the future, I suggest you ditch Windows for Linux or OS X.