Cas9 targets genomic loci with high specificity. For knockin with double-strand break repair, however, Cas9 often leads to unintended on-target knockout rather than intended edits. This imprecision is a barrier for direct in vivo editing where clonal selection is not feasible. In this study, we demonstrate a high-throughput workflow to comparatively assess on-target efficiency and precision of editing outcomes. Using this workflow, we screened combinations of donor DNA and Cas9 variants, as well as fusions to DNA repair proteins. This yielded novel high-performance double-strand break repair editing agents and combinatorial optimizations, yielding increases in knockin efficiency and precision. Cas9-RC, a novel fusion Cas9 flanked by eRad18 and CtIP[HE], increased knockin performance in vitro and in vivo in the developing mouse brain. Continued comparative assessment of editing efficiency and precision with this framework will further the development of high-performance editing agents for in vivo knockin and future genome therapeutics.