v4.3.0

Key Feature / 关键更新

🧩 Production-grade Plugin System / 生产级插件系统

我们重新设计并实现了插件系统，新插件系统采用独立进程+WebSocket通信的模式，隔离插件代码与 LangBot 主程序代码，提供高安全性和灵活性。同时，langbot-plugin-sdk提供了一套完善的 LangBot CLI 工具辅助插件开发，并提供统一、丰富的插件 API 和实体引用。全套插件系统均支持多种语言，并对扩展开放。

We've redesigned and reimplemented the plugin system. The new system uses a standalone process with WebSocket communication, isolating plugin code from the main LangBot program code and providing high security and flexibility. Furthermore, langbot-plugin-sdk provides a comprehensive LangBot CLI tool to assist with plugin development, along with a unified and rich plugin API and entity references. The entire plugin system supports multiple languages ​​and is open to extensions.

我们开放了首个中心化服务：LangBot Space，此服务现提供插件市场功能，您可从此处安装插件，或使用 LangBot CLI 工具上传插件到市场。

We have opened our first centralized service: LangBot Space. This service now provides a Plugin Marketplace feature. You can install plugins from here or upload plugins to the market using the LangBot CLI tool.

• 新版插件机制详细文档请见：插件系统

• 插件注册方式和发行方式已重构，以往的插件开发者请尽快迁移

• 链接：插件市场, 插件 SDK 仓库, 插件 Demo 仓库

• For detailed documentation on the new plugin system, please see: Plugin System

• Plugin registration and distribution have been restructured. Existing plugin developers are advised to migrate as soon as possible

• Links: Plugin Market, Plugin SDK Repository, Plugin Demo Repository

👋 New Adapters / 新适配器

LINEユーザーの皆さん、こんにちは！LangBot 现在支持 LINE 机器人，欢迎尝试。

LangBot now supports LINE bots, welcome to try.

同时添加了备受期待的企业微信智能机器人支持，现在已可使用。

What's Changed

• Fix: Fixed the incorrect extraction method of sender ID when converting aiocqhttp reply messages by @Ljzd-PRO in #1624
• Update package.json by @TwperBody in #1627
• feat: add ZIP file upload support for knowledge base by @devin-ai-integration[bot] in #1626
• fix dingtalk message sender id & update dingtalk streaming card without content by @0xzmz in #1630
• add cmd enable config & fix announce request timeout & fix send card with disconnect ai platform by @0xzmz in #1633
• feat: add GitHub star count component to sidebar by @devin-ai-integration[bot] in #1636
• feat: supported langflow api provider by @ozshen in #1646
• Feat/qdrant vdb by @wangcham in #1649
• chore(deps): bump axios from 1.9.0 to 1.12.0 in /web by @dependabot[bot] in #1655
• chore(deps): bump next from 15.2.4 to 15.4.7 in /web by @dependabot[bot] in #1656
• feat: new plugin system by @RockChinQ in #1495
• Rc/new plugin by @RockChinQ in #1659
• add database connect config by @0xzmz in #1647
• Feat/wecom ai bot by @wangcham in #1654
• fix: wecom function by @wangcham in #1664
• Feat/lineadapter by @fdc310 in #1637

New Contributors

• @0xzmz made their first contribution in #1630
• @ozshen made their first contribution in #1646

更新方式

• Docker：docker-compose.yaml 文件已移动到 docker/ 目录下，请拉取最新代码并按照文档启动
• 手动部署：请按照文档使用

Full Changelog: v4.2.1...v4.3.0

v4.2.2

What's Changed

• Fix: Fixed the incorrect extraction method of sender ID when converting aiocqhttp reply messages by @Ljzd-PRO in #1624
• feat: add ZIP file upload support for knowledge base by @devin-ai-integration[bot] in #1626
• fix dingtalk message sender id & update dingtalk streaming card without content by @0xzmz in #1630
• add cmd enable config & fix announce request timeout & fix send card with disconnect ai platform by @0xzmz in #1633
• feat: add GitHub star count component to sidebar by @devin-ai-integration[bot] in #1636

New Contributors

• @0xzmz made their first contribution in #1630

Full Changelog: v4.2.1...v4.2.2

v4.3.0.beta4

What's Changed

• Update package.json by @TwperBody in #1627
• feat: add ZIP file upload support for knowledge base by @devin-ai-integration[bot] in #1626
• Merge commits from master and fix bugs

Full Changelog: v4.3.0.beta2...v4.3.0.beta4

v4.3.0.beta3

Full Changelog: v4.3.0.beta1...v4.3.0.beta3

v4.3.0.beta2

What's Changed

• Fix: Fixed the incorrect extraction method of sender ID when converting aiocqhttp reply messages by @Ljzd-PRO in #1624

Full Changelog: v4.2.1...v4.3.0.beta2

v4.3.0.beta1

Full Changelog: v4.0.7...v4.3.0.beta1

v4.2.1

What's Changed

• fix:in the dify agent llm return message can not joint by @fdc310 in #1617
• Fix: Fixed the issue where Ollama's embedding model was not functioning properly by @Ljzd-PRO in #1619
• fix:in the gmini tool_calls no id The resulting call failure by @fdc310 in #1622

New Contributors

• @Ljzd-PRO made their first contribution in #1619

Full Changelog: v4.2.0...v4.2.1

v4.2.0

Key Feature / 关键更新

🥷 Streaming Output / 流式输出

由 @fdc310 贡献；现在 LangBot 的模型、流水线、机器人（视平台而定）已适配流式输出特性，仅需在适配器开启流式输出，即可获得 AI 流式输出。并已在流水线内置思维链移除功能。

Contributed by @fdc310; LangBot's models, pipelines, and robots (depending on the platform) are now compatible with streaming output. Simply enable streaming output in the adapter to get AI streaming output. A thought chain removal function has also been built into the pipeline.

• 目前支持的平台：飞书、钉钉、Telegram、WebUI 调试会话。
• Currently supports: Telegram, Lark, DingTalk, WebUI

🌒 Dark Mode / 暗色模式

由 @TwperBody 贡献；
Contributed by @TwperBody

What's Changed

• fix(pipeline dialog): config reset between tabs switching by @RockChinQ in #1597
• feat: dark mode supports for webui by @TwperBody in #1605
• Feat/topk splitter by @wangcham in #1606
• feat: add password change functionality by @devin-ai-integration[bot] in #1610
• feat:add streaming output and pipeline stream by @fdc310 in #1571
• feat: streaming output by @RockChinQ in #1442
• chore(deps): bump form-data from 4.0.2 to 4.0.4 in /web by @dependabot[bot] in #1612

Full Changelog: v4.1.2...v4.2.0

v4.1.2

What's Changed

• feat: add new api requester by @RockChinQ in #1596

Full Changelog: v4.1.1...v4.1.2

v4.1.1

What's Changed

• feat: add pipeline sorting functionality with three sort options by @devin-ai-integration[bot] in #1582
• feat: add knowledge base retrieve test tab with Card-based UI by @devin-ai-integration[bot] in #1583
• feat(wechatpad): 添加对@所有人的支持并统一处理消息派发 by @gaord in #1588
• Add GitHub link redirection for front-end plugin cards by @TwperBody in #1579

Full Changelog: v4.1.0...v4.1.1