## 更新内容
* chore: 将版本提升至 0.8.2 by @Mingholy in https://github.com/QwenLM/qwen-code/pull/1632
* chore(release): sdk-typescript v0.1.4 by @qwen-code-ci-bot in https://github.com/QwenLM/qwen-code/pull/1633
* Fix: 使用解析后的 authType 初始化 ACP agent by @Mingholy in https://github.com/QwenLM/qwen-code/pull/1622
* 为 ACP 子代理追踪添加 parentToolCallId 和 subagentType by @Mingholy in https://github.com/QwenLM/qwen-code/pull/1620
* feat(core): 改进以工具调用 skill 时的错误提示 by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1623
* fix(acp): 流式输出子代理文本与推理块 by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1626
* fix: 确保在配置初始化前创建 output-language.md by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1637
* fix(security): 防止在 shell 命令验证中通过换行符绕过导致命令注入 by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1638
* Fix: 使 Claude 插件资源收集遵循 marketplace 配置 by @LaZzyMan in https://github.com/QwenLM/qwen-code/pull/1639
* feat(core): 改进重试逻辑以更好地处理 429/5xx 错误 by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1628
* feat: 为 OpenAI 兼容提供商添加 extra_body 支持 by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1654
* 在 vscode ide companion 中支持通过 ACP 进行模型选择 by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/1582
* feat: 在所有内容生成器中增加多模态输入支持（图像、PDF、音频） by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1564
* 为 Qwen Code agent server 添加 Zed 扩展 by @Mingholy in https://github.com/QwenLM/qwen-code/pull/1630
* feat: 明确非交互模式的输出格式 by @LaZzyMan in https://github.com/QwenLM/qwen-code/pull/1579
* feat: 为批量 CLI 执行添加并发运行器 by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1640
* 为代码智能添加实验性 LSP 支持 by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/1401
* Fix: 修复 package.json 和 lockfile 中 React/React-DOM 版本不一致问题 by @Mingholy in https://github.com/QwenLM/qwen-code/pull/1659
* feat(webui): 使用共享组件库实现统一的 UI 架构 by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/1543
* chore: 将版本提升至 0.9.0 by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1661
* 添加 contextWindowSize 配置支持 by @xuewenjie123 in https://github.com/QwenLM/qwen-code/pull/1539
* fix(core): 避免将 undici agent 传递给 Anthropic SDK by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1663
* fix(vscode-ide-companion): 修复竞态条件并优化 @ 文件补全搜索 by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/1676
**完整更新日志**: https://github.com/QwenLM/qwen-code/compare/v0.8.2...v0.9.0
## What's Changed
* chore: bump version to 0.8.2 by @Mingholy in https://github.com/QwenLM/qwen-code/pull/1632
* chore(release): sdk-typescript v0.1.4 by @qwen-code-ci-bot in https://github.com/QwenLM/qwen-code/pull/1633
* Fix: Use resolved authType to initialize ACP agent by @Mingholy in https://github.com/QwenLM/qwen-code/pull/1622
* Add parentToolCallId and subagentType for ACP subagent tracking by @Mingholy in https://github.com/QwenLM/qwen-code/pull/1620
* feat(core): improve error message when skill is invoked as tool by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1623
* fix(acp): stream subagent text + reasoning chunks by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1626
* fix: ensure output-language.md is created before config initialization by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1637
* fix(security): prevent command injection via newline bypass in shell command validation by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1638
* Fix Claude plugin resource collection to respect marketplace config by @LaZzyMan in https://github.com/QwenLM/qwen-code/pull/1639
* feat(core): improve retry logic for better 429/5xx error handling by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1628
* feat: add extra_body support for OpenAI-compatible providers by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1654
* Support model selection through ACP in vscode ide companion by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/1582
* feat: add multi-modal input support (image, PDF, audio) across all content generators by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1564
* Add Zed extension for Qwen Code agent server by @Mingholy in https://github.com/QwenLM/qwen-code/pull/1630
* feat: clarify output formats for non-interactive mode by @LaZzyMan in https://github.com/QwenLM/qwen-code/pull/1579
* feat: add concurrent runner for batch CLI execution by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1640
* Add experimental LSP support for code intelligence by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/1401
* Fix: React/React-DOM version inconsistency in package.json and lockfile by @Mingholy in https://github.com/QwenLM/qwen-code/pull/1659
* feat(webui): implement unified UI architecture with shared component library by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/1543
* chore: bump version to 0.9.0 by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1661
* Add contextWindowSize Configuration Support by @xuewenjie123 in https://github.com/QwenLM/qwen-code/pull/1539
* fix(core): avoid passing undici agent to Anthropic SDK by @tanzhenxin in https://github.com/QwenLM/qwen-code/pull/1663
* fix(vscode-ide-companion): fix race conditions and improve @ file completion search by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/1676
**Full Changelog**: https://github.com/QwenLM/qwen-code/compare/v0.8.2...v0.9.0