Woohoo!!! 
!!!
With that little bit of data you provided, I was able to reproduce the error. Not only that, but I was able to see (
) exactly what the issue is. Running the same command you provided above, but instead of using q2-cutadapt we go through cutadapt directly, I get some reads out of cutadapt that look like this:
Yowzah! So, hopefully that QIIME 2 error makes a bit more sense now --- it was complaining about records that had no sequence associated with it.
I have opened up an issue on the relevant bug tracker for us to fix this up. In the meantime, you could export your demuxed reads and run cutadapt on its own, then reimport your trimmed reads, just make sure to include the flag -m 1 to specify a minimum read length (you could use a larger value, but don't set it any lower than this. For example:
cutadapt \
--cores 1 \
--error-rate 0.1 \
--times 1 \
--overlap 3 \
-o trimmed/Teleo_1_L001_R1_001.fastq.gz \
--adapter CTTCCGGTACACTTACCATG \
--front ACACCGCCCGTCACTCT \
-m 1 \
input/Teleo_1_L001_R1_001.fastq.gz
Thanks so much for your help with identifying this bug - you saved the day! 