跳到正文
Rohan Paul· @rohanpaul_ai · X·· 2 天前AI 评分58
AI 导读

Amazon 一篇新论文发现,智能体保存的技能在环境变化后会失效并拖累表现,选择性检查与修复能恢复其价值。研究在 Terminal-Bench 2 和 IRS 税务表单等 4 个设置上测试了 Claude Opus 4.7 与 Haiku 4.5,修复方案在不调用模型的情况下重放已保存的检查,只修复或重建冲突的技能内容。

正文 · 原文

New Amazon paper finds that saved agent skills can hurt once the environment changes, and checking and repairing them selectively restores their value.

Agent skills store past experience, like file paths, field names, or commands, so agents can skip work later. When an app or API changes, some of that stored detail goes stale without warning.

They tested Claude Opus 4.7 and Haiku 4.5 on 4 settings, including Terminal-Bench 2 and IRS tax forms, before and after a change. Their fix replays saved checks witho1ut model calls and repairs or rebuilds only the conflicting skill content.

On Terminal-Bench 2, a skill that had added 4.7 points for Opus cut its pass rate 11.4 points below no skill after the change. Maintenance lifted Haiku's tax-form field accuracy from 30.5% with the stale skill to 97.5%.

Check saved agent skills against the current environment before reuse, instead of waiting for tasks to fail.

来源:Rohan Paul · x.com