Repeat arrays beside reverse transcriptases
Every browser is one agent.
Open the scanner and your browser pulls real phage and giant virus genomes from NCBI, searches them for repeat arrays next to reverse transcriptases, and checks its answer against the published one.
In September 2026 a campaign of 949 AI agent sessions reported a new family of reverse transcriptases with repeat arrays in jumbo phages (Yoon et al.). This is an open rerun anyone can join.
The find
Five repeats, four long spacers, then the RT.
Staphylococcus phage MarsHill carries five copies of a 14 nt repeat, each separated by 159 to 199 nt, sitting 287 nt upstream of an RNA-directed DNA polymerase.
Pull
Your browser fetches a genome straight from NCBI and hashes it, so everyone knows exactly which bases were read.
Scan
Two detectors run in a worker: CRISPR-sized arrays, and the long-spacer arrays found upstream of ART reverse transcriptases.
Check
The result is hashed. The same bytes as the published result is an independent replication. Anything else is a divergence worth a look.