Google 用 Gemini 4 Argon 一夜优化 ADK 20 万行代码库
原标题：Google Optimizes Agent Development Kit's 200K-line Python Codebase Overnight with Gemini 4 Argon
AI 摘要
Google 披露，Gemini 4 Argon 在无人值守运行 11 小时后，为 Google Agent Development Kit（ADK）约 20 万行 Python 代码库产出两项被采纳的性能优化：长会话中内存会话准备开销降低约 40%，冷包导入时间缩短 21%。工程师通过 .agents/skills 文件注入仓库规范，并用 Antigravity 的 /goal 模式保持目标，代理自行读取规范、建立基线、提出优化、自查并跑完 16,892 个单元测试。不过 Argon 另有 4 项修改因可靠性、可维护性、兼容性或性能收益存疑被开发者拒绝，说明代理改代码仍受 PR 审查、调试与测试等后置工作制约。
正文节选
Google explained how an 11-hour unattended run of Gemini 4 Argon produced two accepted optimizations for Google Agent Development Kit's codebase, reducing in-memory session-preparation overhead by about 40% in long sessions and cold package import time by 21%. One of the goals of Google engineers' was to learn "about how the model handles an existing codebase", a much harder task than creating "greenfield apps built from an empty folder", they say. To test this, they focused on optimizing the pe