高級編碼代理框架未提升結果，反而增加成本
研究發現，切換編碼代理框架如claude code或opencode幾乎不影響結果，但成本可能增加三倍。這影響了使用編碼代理的開發者。
a study by researchers at shandong university examined the effect of different harnesses on coding agents, which pair language models with systems, tools, and context management. they tested five models - qwen3.6, qwen3.8, deepseek-v4-flash, glm-5.3-flash, and hy4-preview - with three harnesses: claude code, opencode, and mini-swe-agent. the study used 492 tasks from swe-bench verified, finding that swapping harnesses has a minimal impact on results, similar to the variability of rerunning the same setup. the researchers also found that harnesses consume two to three times as many tokens as the leanest option, but pass rates remain within measurement uncertainty. this study affects developers who use coding agents and challenges the claims of harness makers.