Since you don't mention adapter content problem in your samples then trimming is unnecessary. Duplicates and overrepresented k-mers are expected in RNAseq data. Uneven GC content is also very common in such libraries. I think you have nothing to be worried about if you have good quality reads. Carry on the processing and make you get a nice mapping to the reference genome.
About the sequence duplication levels and GC content, sometimes they reflect the characteristics of species you have used (e.g duplication rate is up in most fish and plants), and about the K-mer content sometimes it is because of barcodes/adapters you have used in sequencing procedures or poor libraries.
For trimming (if needed) you can use FASTX (which is very easy to use) or Trimmomatic software (which the later one is embedded in some assembler same as Trinity, too).
Additionaly you can check the "Common reasons for warnings" part of each title in this page.
If you are using Galaxy, you can find some useful tools under "NGS: QC and manipulation".
By the way, remember to keep a backup of your data when begin to trimming them :-)