Repowise scores nousresearch/hermes-agent at 5.3 out of 10 for code health, which reads as needs work, alongside maintainability 6.8 and static performance 9.4 on the same scale. There are 43,909 open findings across the three pillars, and the lowest-scoring file is run.py at 1.0. Scores are computed from complexity, duplication, churn and git fix history, not from a linter.