Archived
This repository has been archived on 2026-09-13 . You can view files and clone it. You cannot open issues or pull requests or push a commit.
154b2228609b7bca5a82ee613a5100931568f80b
R10:评审查实 Task 12 只测了 actorOf 这个纯函数,而 auto-runner.ts 四类留痕
(say / project / analyze / 出错兜底)一直用 kind: 'result',它不在 task-log.ts 的
SKELETON 里,structuredLog 在骨架过滤这一关就把它们全部滤掉,actorOf 判得再对也没用。
project 与 analyze 两笔还带 detail.internal: true,即使 kind 改对也会被 machinery() 挡下,
一并去掉。
新增回归测试:直接跑真实的 auto-runner.ts,把捕获到的 {kind, title, detail} 喂给
structuredLog 全程,断言它出现在结果里且 who 是 system —— 不是照 spec 手写理想形状。
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LKMZDwMbyKzjw6Hembwvet
Description
No description provided
9.3 MiB
Languages
TypeScript
90.1%
CSS
7.3%
JavaScript
2%
Shell
0.4%
Python
0.1%