## 更新内容
* chore(release): 发布 v0.15.6 by @qwen-code-ci-bot in https://github.com/QwenLM/qwen-code/pull/3766
* feat(core): 添加 FileReadCache 并短路跳过未更改的读取操作 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3717
* fix(cli): 遵循代理设置 by @cyphercodes in https://github.com/QwenLM/qwen-code/pull/3753
* chore: 移除旧的 Gemini 工作流 by @pomelo-nwu in https://github.com/QwenLM/qwen-code/pull/3725
* feat(core): 添加工具执行统一所需的共享权限流程 by @B-A-M-N in https://github.com/QwenLM/qwen-code/pull/3723
* fix(cli): 恢复 SubAgent 快捷键焦点 by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/3771
* 添加后台代理的恢复与继续功能 by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3739
* feat(review): 扩展审查流水线 + 新增 qwen review CLI 子命令 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3754
* feat(telemetry): 定义 HTTP OTLP 端点行为与信号路由 by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3779
* fix(vscode-companion): 使包级 eslint 配置与根目录对齐并清理样式 by @B-A-M-N in https://github.com/QwenLM/qwen-code/pull/3782
* feat(core): 带节流 stdout 流的事件监控工具（Phase C） by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3684
* fix(test): 将 abort-and-lifecycle stdin-close 测试恢复至 #3723 之前版本 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3777
* feat(cli): 在页脚添加 MCP 健康状态指示器 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3741
* fix(core): 为 DeepSeek anthropic 兼容提供商注入思考块 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3788
* Feat/stats 模型成本估算变基 by @B-A-M-N in https://github.com/QwenLM/qwen-code/pull/3780
* fix(cli): 在非交互模式下停止 API 错误的重复包装与重复打印 by @umut-polat in https://github.com/QwenLM/qwen-code/pull/3749
* feat(cli): 将 Monitor 条目接入组合后台任务对话框 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3791
* feat(cli): 在 /tasks 中包含监控项 + 添加交互模式提示 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3801
* fix(telemetry): 启动时抑制异步资源属性警告 by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3807
* feat(sdk-python): 添加 PyPI 发布工作流 by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3685
* fix(core): 解决合并后的监控工具与 UI 路由问题 by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3792
* docs(core): 将 background-shell + monitor 指引同时指向 /tasks 和对话框 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3808
* fix(core): 在每次历史重写路径中清除 FileReadCache by @wenshao in https://github.com/QwenLM/qwen-code/pull/3810
* feat(core): 支持推理强度 'max' 级别（DeepSeek 扩展） by @wenshao in https://github.com/QwenLM/qwen-code/pull/3800
* feat(core): 提示后台长耗时前台 bash 命令 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3809
* feat(skills): 并行加载 + 添加路径条件激活 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3604
* fix(core): 在 Edit、WriteFile 和 ReadFile 工具中还原 shell 转义的文件路径 by @qiuqiuwen25 in https://github.com/QwenLM/qwen-code/pull/3820
* fix(openai): 解析 MiniMax 思考标签 by @shenyankm in https://github.com/QwenLM/qwen-code/pull/3677
* refactor: 提取共享的发布辅助工具 by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3834
* fix(telemetry): 添加有界关闭超时并修复 service.version 资源属性 by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3813
* fix(acp): 在模型发送前运行自动压缩 by @Jerry2003826 in https://github.com/QwenLM/qwen-code/pull/3698
* fix(core): 合并 MCP 服务器重新发现逻辑 by @cyphercodes in https://github.com/QwenLM/qwen-code/pull/3818
* feat(sdk-python): 为发布版本辅助工具添加网络超时 by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3833
* feat(cli): 改进导出格式补全导航 by @shenyankm in https://github.com/QwenLM/qwen-code/pull/3701
* feat(cli): 支持通过 CLI 非交互式切换模型 by @alex-musick in https://github.com/QwenLM/qwen-code/pull/3783
* feat(weixin): 通过 CDN 上传添加图片发送支持 by @Mr-Maidong in https://github.com/QwenLM/qwen-code/pull/3781
* fix(core): 从发现的结果路径激活技能 by @qiuqiuwen25 in https://github.com/QwenLM/qwen-code/pull/3852
* fix(core): 使用按模型设置进行快速模型侧查询 by @B-A-M-N in https://github.com/QwenLM/qwen-code/pull/3815
* fix(core): 防止自动记忆召回阻塞主请求 by @B-A-M-N in https://github.com/QwenLM/qwen-code/pull/3814
* feat(core,cli): 展示并取消自动记忆梦境任务 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3836
* fix(sdk-python): 标准化 TAG_PREFIX 以包含 v 后缀 by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3832
* fix(cli): 防止文件路径被误识别为斜杠命令 by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/3743
* feat(cli): 运行时通过指示器+对话框路由前台子代理 by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/3768
* fix(core): 自动压缩子代理上下文以防止溢出 by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/3735
* feat(core): 在 Edit / WriteFile 修改文件前强制先读取 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3774
* fix(core): 缩减文件 diff 会话记录 by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3872
* feat(cli): 自定义横幅区域（logo、标题、隐藏） by @chiga0 in https://github.com/QwenLM/qwen-code/pull/3710
* feat(core): 为 ShellExecutionService 添加 signal.reason 约定 (#3831 PR-1 of 3) by @wenshao in https://github.com/QwenLM/qwen-code/pull/3842
* fix(core): 在子代理配置覆盖时重建工具注册表，使绑定的工具正确解析至子代理 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3873
* fix(core): 在保存截断的 shell 输出前创建临时目录 by @kkhomej33-netizen in https://github.com/QwenLM/qwen-code/pull/3875
* fix(core): 改进流式速率限制的重试处理机制 by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/3790
* docs(cli): 新增横幅自定义设置的文档说明 by @wenshao in https://github.com/QwenLM/qwen-code/pull/3885
* fix(core): 处理 @tanzhenxin 的 PR-1 评审意见（#3842 合并后的跟进修复） by @wenshao in https://github.com/QwenLM/qwen-code/pull/3886
* fix(core): 在前台 fork 路径上停止各子代理的 ToolRegistry by @wenshao in https://github.com/QwenLM/qwen-code/pull/3887
* ci: 新增 Qwen Code issue 跟进机器人工作流 by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/3854
## 新贡献者
* @umut-polat 在 https://github.com/QwenLM/qwen-code/pull/3749 完成首次贡献
* @qiuqiuwen25 在 https://github.com/QwenLM/qwen-code/pull/3820 完成首次贡献
* @alex-musick 在 https://github.com/QwenLM/qwen-code/pull/3783 完成首次贡献
* @Mr-Maidong 在 https://github.com/QwenLM/qwen-code/pull/3781 完成首次贡献
**完整更新日志**: https://github.com/QwenLM/qwen-code/compare/v0.15.6...v0.15.7-preview.2
## What's Changed
* chore(release): v0.15.6 by @qwen-code-ci-bot in https://github.com/QwenLM/qwen-code/pull/3766
* feat(core): add FileReadCache and short-circuit unchanged Reads by @wenshao in https://github.com/QwenLM/qwen-code/pull/3717
* fix(cli): honor proxy setting by @cyphercodes in https://github.com/QwenLM/qwen-code/pull/3753
* chore: remove legacy Gemini workflows by @pomelo-nwu in https://github.com/QwenLM/qwen-code/pull/3725
* feat(core): add shared permission flow for tool execution unification by @B-A-M-N in https://github.com/QwenLM/qwen-code/pull/3723
* fix(cli): restore SubAgent shortcut focus by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/3771
* Add background agent resume and continuation by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3739
* feat(review): expand review pipeline + qwen review CLI subcommands by @wenshao in https://github.com/QwenLM/qwen-code/pull/3754
* feat(telemetry): define HTTP OTLP endpoint behavior and signal routing by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3779
* fix(vscode-companion): align package eslint config with root and style cleanup by @B-A-M-N in https://github.com/QwenLM/qwen-code/pull/3782
* feat(core): event monitor tool with throttled stdout streaming (Phase C) by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3684
* fix(test): restore abort-and-lifecycle stdin-close test to pre-#3723 version by @wenshao in https://github.com/QwenLM/qwen-code/pull/3777
* feat(cli): add MCP health pill to footer by @wenshao in https://github.com/QwenLM/qwen-code/pull/3741
* fix(core): inject thinking blocks for DeepSeek anthropic-compatible provider by @wenshao in https://github.com/QwenLM/qwen-code/pull/3788
* Feat/stats model cost estimation rebase by @B-A-M-N in https://github.com/QwenLM/qwen-code/pull/3780
* fix(cli): stop double-wrapping and double-printing API errors in non-interactive mode by @umut-polat in https://github.com/QwenLM/qwen-code/pull/3749
* feat(cli): wire Monitor entries into combined Background tasks dialog by @wenshao in https://github.com/QwenLM/qwen-code/pull/3791
* feat(cli): include monitors in /tasks + add interactive-mode hint by @wenshao in https://github.com/QwenLM/qwen-code/pull/3801
* fix(telemetry): suppress async resource attribute warning on startup by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3807
* feat(sdk-python): add PyPI release workflow by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3685
* fix(core): address post-merge monitor tool and UI routing issues by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3792
* docs(core): point background-shell + monitor guidance at both /tasks and the dialog by @wenshao in https://github.com/QwenLM/qwen-code/pull/3808
* fix(core): clear FileReadCache on every history rewrite path by @wenshao in https://github.com/QwenLM/qwen-code/pull/3810
* feat(core): support reasoning effort 'max' tier (DeepSeek extension) by @wenshao in https://github.com/QwenLM/qwen-code/pull/3800
* feat(core): hint to background long-running foreground bash commands by @wenshao in https://github.com/QwenLM/qwen-code/pull/3809
* feat(skills): parallelize loading + add path-conditional activation by @wenshao in https://github.com/QwenLM/qwen-code/pull/3604
* fix(core): unescape shell-escaped file paths in Edit, WriteFile, and ReadFile tools by @qiuqiuwen25 in https://github.com/QwenLM/qwen-code/pull/3820
* fix(openai): parse MiniMax thinking tags by @shenyankm in https://github.com/QwenLM/qwen-code/pull/3677
* refactor: extract shared release helper utilities by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3834
* fix(telemetry): add bounded shutdown timeout and fix service.version resource attribute by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3813
* fix(acp): run auto compression before model sends by @Jerry2003826 in https://github.com/QwenLM/qwen-code/pull/3698
* fix(core): coalesce MCP server rediscovery by @cyphercodes in https://github.com/QwenLM/qwen-code/pull/3818
* feat(sdk-python): add network timeouts to release version helper by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3833
* feat(cli): improve export format completion navigation by @shenyankm in https://github.com/QwenLM/qwen-code/pull/3701
* feat(cli): Add ability to switch models non-interactively from the cli by @alex-musick in https://github.com/QwenLM/qwen-code/pull/3783
* feat(weixin): add image sending support via CDN upload by @Mr-Maidong in https://github.com/QwenLM/qwen-code/pull/3781
* fix(core): activate skills from discovered result paths by @qiuqiuwen25 in https://github.com/QwenLM/qwen-code/pull/3852
* fix(core): use per-model settings for fast model side queries by @B-A-M-N in https://github.com/QwenLM/qwen-code/pull/3815
* fix(core): prevent auto-memory recall from blocking main request by @B-A-M-N in https://github.com/QwenLM/qwen-code/pull/3814
* feat(core,cli): surface and cancel auto-memory dream tasks by @wenshao in https://github.com/QwenLM/qwen-code/pull/3836
* fix(sdk-python): standardize TAG_PREFIX to include v suffix by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3832
* fix(cli): prevent file paths from being treated as slash commands by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/3743
* feat(cli): route foreground subagents through pill+dialog while running by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/3768
* fix(core): auto-compact subagent context to prevent overflow by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/3735
* feat(core): enforce prior read before Edit / WriteFile mutates a file by @wenshao in https://github.com/QwenLM/qwen-code/pull/3774
* fix(core): shrink file diff session records by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/3872
* feat(cli): customize banner area (logo, title, hide) by @chiga0 in https://github.com/QwenLM/qwen-code/pull/3710
* feat(core): add signal.reason convention for ShellExecutionService (#3831 PR-1 of 3) by @wenshao in https://github.com/QwenLM/qwen-code/pull/3842
* fix(core): rebuild tool registry on subagent Config overrides so bound tools resolve to the subagent by @wenshao in https://github.com/QwenLM/qwen-code/pull/3873
* fix(core): create temp dir before saving truncated shell output by @kkhomej33-netizen in https://github.com/QwenLM/qwen-code/pull/3875
* fix(core): improve stream rate-limit retry handling by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/3790
* docs(cli): document new banner customization settings by @wenshao in https://github.com/QwenLM/qwen-code/pull/3885
* fix(core): address @tanzhenxin's PR-1 review notes (post-merge follow-up to #3842) by @wenshao in https://github.com/QwenLM/qwen-code/pull/3886
* fix(core): stop per-subagent ToolRegistry on foreground-fork path by @wenshao in https://github.com/QwenLM/qwen-code/pull/3887
* ci: add Qwen Code issue follow-up bot workflow by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/3854
## New Contributors
* @umut-polat made their first contribution in https://github.com/QwenLM/qwen-code/pull/3749
* @qiuqiuwen25 made their first contribution in https://github.com/QwenLM/qwen-code/pull/3820
* @alex-musick made their first contribution in https://github.com/QwenLM/qwen-code/pull/3783
* @Mr-Maidong made their first contribution in https://github.com/QwenLM/qwen-code/pull/3781
**Full Changelog**: https://github.com/QwenLM/qwen-code/compare/v0.15.6...v0.15.7-preview.2