Hacktoberfest 2026:维护者为十月标记出来的 issue,仍然开放、适合新手。 浏览 Hacktoberfest issue

/refine failures

未关闭
#4,747 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

还没有人认领这个 Issue。

评估

难度
3/5
预计耗时
1-2 天
新手友好度
62/100
Issue 类型
缺陷
描述清晰度
基本清楚
活跃度
活跃
技术栈
azure, shell
领域
api, cli

调研方向

从 /refine 命令开始,跟踪其对 gpt-4o-mini 的内部请求,重点关注 reasoningEffort 是如何添加的,以及 turn_failed 是如何处理的。完成的标准是请求不再发送不受支持的设置,并且显示实际的失败,而不是静默地报告文本未发生变化。

由索引模型根据 Issue 内容生成。

描述

triage
Describe the bug

Session Diagnosis

Root cause found. At  06:55:33Z , the  /refine  command failed with a 400 Bad Request:

"reasoning_effort "medium" was provided, but model gpt-4o-mini-2024-07-18 does not support reasoning effort"

What happened:

  1.  /refine  internally invokes  gpt-4o-mini  (a lightweight utility model) to rewrite your prompt into    tags.
  2. The request incorrectly included  reasoningEffort: "medium"  as a parameter.
  3.  gpt-4o-mini-2024-07-18  doesn't support the  reasoning_effort  parameter at all (only reasoning-capable models like o1/o3 do) — Azure OpenAI rejected the call outright with  invalid_reasoning_effort .
  4. The call failed in 545ms,  turn_failed  fired, and the CLI silently fell back to "Your text was left unchanged" instead of surfacing the real error.

This is a CLI-side bug, not an environment/proxy/auth issue: the refine feature is passing a reasoning-effort setting to a model that categorically rejects it. It's unrelated to the Blue Coat proxy or credentials — the request reached the API fine and got a clean structured rejection.

Affected version

1.0.83

Steps to reproduce the behavior

No response

Expected behavior

No response

Additional context

No response

主要语言
Shell
星标
11.2k
派生
1.9k
平均合并
14 小时 16 分钟
30 天内合并 PR
6

贡献指南

打开贡献指南

从这里开始

  1. 先读完整个 Issue,再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库,在一个分支上完成修改。
  4. 提交 Pull Request,并在描述里引用这个 Issue 编号。

github/copilot-cli 的其他 Issue

查看 github/copilot-cli 的全部 Issue

相似的 Issue

更多 Shell/Bash Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。