Hepatitis A virus (HAV) infections are an increasing public health concern in low-endemicity regions due to outbreaks from foodborne infections and sustained transmission among vulnerable groups, including persons experiencing homelessness, those who inject drugs, and men who have sex with men (MSM), which is further compounded by aging, unvaccinated populations. DNA sequence characterisation of HAV for source tracking is performed by comparing small subgenomic regions of the virus. While this approach has been successful when robust epidemiological data are available, poor genetic resolution can lead to conflation of outbreaks with sporadic cases. HAV outbreak investigations would greatly benefit from the additional phylogenetic resolution obtained by whole virus genome sequence comparisons. However, HAV genomic approaches can be difficult because of challenges in isolating the virus, low sensitivity of direct metagenomic sequencing in complex sample matrices like various foods such as fruits, vegetables and molluscs, and difficulty designing highly multiplexed PCR primers across diverse HAV genotypes. Here, we introduce a proof-of-concept pan-HAV oligonucleotide hybrid capture enrichment assay from serum and frozen berry specimens that yields complete and near-complete HAV genomes from as few as four input HAV genome copies. We used this method to recover HAV genomes from human serum specimens with high CT values (34.7-42.7), with high assay performance for all six human HAV sub-genotypes, both contemporary and historical. Our approach provides a highly sensitive and streamlined workflow for HAV WGS from diverse sample types, that can be the basis for harmonised and high-resolution molecular epidemiology during HAV outbreak surveillance.