We revisit the evaluation of automatic harness evolution for LLM agents. Existing harness evolution methods use unit test cases to search for harness configurations and report final performance on the same public benchmark. This protocol raises two fundamental concerns.
First, harness evolution is itself an iterative search procedure that repeatedly evaluates and revises candidate harnesses using task feedback. Therefore, it should be compared with simple task-level search baselines under matched feedback and inference budgets to determine whether gains arise from improved harness design or from additional search alone.
Second, because the search and final evaluation share the same benchmark, reported gains risk overfitting to that specific task set. To address these concerns, we conduct an extensive evaluation comparing harness evolution with simple test-time scaling and discovery baselines under comparable feedback and inference budgets, and also evaluate evolved harnesses on held-out tasks to assess whether discovered improvements generalize.
Experiments on Terminal-Bench 2.1 with GPT-5.4 and Claude Opus 4.6 show that automatic harness evolution does not consistently outperform simple test-time scaling methods and exhibits limited generalization. Our results raise important questions about the effectiveness of automatic harness evolution and highlight the need for fairer evaluation protocols and benchmarks for automatic harness design. Our code is available at GitHub.
Blogger's Review: This article presents significant reflections on the evaluation of automatic harness evolution, particularly regarding overfitting and evaluation protocols. While automation has potential, the experimental results indicate its effectiveness in practical applications requires further verification, reminding us to be cautious in methodological choices when designing and evaluating agents.