anthropics / anthropics/claude-code

[Bug] Claude refuses to modify configuration settings despite explicit user authorization

未关闭
#91,609 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看
area:model area:permissions bug platform:macos
主要语言
Python
星标
145k
派生
23.1k
PR 合并指标
PR 指标待抓取

描述

**Bug Description**
I explicitly asked for a configuration setting change to be written, and it flatly refused even though I fully authorized it and told it that I would approve the right. It absolutely shat in my face. The only thing that got it to actually do it is when I had a fatal model tell it to do it, and then it listened and it actually did what it was meant to do. How is that accurate? How is that helpful? I wasted thirty minutes dealing with this shit. That is ridiculous. If I ask it to write a setting on my machine and I tell it that I'm going to approve it, there is no reason why it should deny me. It should tell me a reason if it's stupid, but it should not flatly deny it.

**Environment Info**
- Platform: darwin
- Terminal: iTerm.app
- Version: 2.1.258
- Feedback ID: 40c2f4a3-07b4-4acc-b962-f289a861025a

**Errors**
```json
[]
```

贡献指南

这个仓库没有索引到贡献指南

调研方向

Start by reproducing the authorized configuration-setting request in Claude Code on darwin using iTerm.app, then compare it with the request that succeeded after the alternate model prompt. Trace the permission or authorization flow involved; done means the behavior is explained and the expected approved-request handling is verified, including any refusal reason.

由索引模型根据 Issue 内容生成。

评估

技术栈
macos, python
领域
authorization, cli
Issue 类型
缺陷
难度
4/5
预计耗时
3-5 天
活跃度
活跃
描述清晰度
需要澄清
新手友好度
28/100

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。