- wip console
- wip console
- wip console
- script package
- wip cloud smart
- wip console
- wip console
- wip console
- wip console
- fix: 文件模糊搜索 (#2409)
- feat: 在SDK中添加createOpencodeTui()函数以支持编程式启动TUI (#2410)
- fix: 工作树文件/内容从不包含补丁
- fix: 改进文件/内容返回
- fix: 网络抓取提示错误 (#2424)
- 更新 index.tsx (#2423)
- tweak(edit): 分离编辑工具错误信息，提供更清晰的指导以避免LLM陷入编辑循环 (#2051)
- fix: 文件读取响应
- fix: 默认使用上次使用的模型 (#2443)
- fix: 命令模型选择 (#2448)
- fix: 若首条消息为shell调用则标题未生成 (#2451)
- fix: 使用gpt-5-nano时的标题生成 (#2473)
- fix: 设置预算思考令牌时的最大输出令牌数 (#2056) 合作者：rekram1-node <
[email protected]>
- wip console
- wip console
- wip console
- script package
- wip cloud smart
- wip console
- wip console
- wip console
- wip console
- fix: file fuzzy search (#2409)
- feat: add createOpencodeTui() function to SDK for programmatic TUI launching (#2410)
- fix: worktree file/content never includes patch
- fix: better file/content return
- fix: webfetch prompt mistake (#2424)
- Update index.tsx (#2423)
- tweak(edit): separate edit tool error message with clearer guidance to avoid llm doom editing loop (#2051)
- fix: file read response
- fix: default to last used model (#2443)
- fix: command model selection (#2448)
- fix: title not generated if first msg is shell invocation (#2451)
- fix: title gen w/ gpt-5-nano (#2473)
- fix: max output tokens when setting budget thinking tokens (#2056) Co-authored-by: rekram1-node <
[email protected]>