💡Features

• Use keyword or fallback search to chat with DeepSeek in alfred:
• Use websearch mode(Fn↩︎) to get up-to-date information:
• Use powerful mode(⌘↩︎) to chat:
• Display reasoning content, if available.
• Interrupted answer will be preserved in the chat history followed by "[Answer Interrupted]".

📝 Prerequisites

You can use alfred-deepseek with one of these providers: DeepSeek official or Volcengine Ark. Currently Volcengine Ark's DeepSeek API's responding speed is much faster, and Volcengine Ark supports websearch mode while DeepSeek official does not. If you want to use DeepSeek via Volcengine Ark, please follow the steps below:

1. Go to https://www.volcengine.com/ , create an account and login:
2. Create a new api key, you can refer to this page https://www.volcengine.com/docs/82379/1399008;
3. Create DeepSeek model endpoints via this page https://console.volcengine.com/ark, say DeepSeek-V3 and DeepSeek-R1:
4. Then click into the model page and click "Deploy and use" button, follow the instructions until finish activating models and creating endpoints:
5. Get model endpoints from the following page: https://console.volcengine.com/ark
6. Ensure you are using Alfred 5.5 or later.

📦 Installation

Just double click the artifact named DeepSeek_by_fanfank_xxxx.alfredworkflow to install.

🔧 Configuration:

Configure the workflow and you are ready to go. 🚀

Don't forget to add DeepSeek to fallback search!

If you want to use WebSearch mode, follow the steps below:

1. Activate ark web search plugin via this page: https://console.volcengine.com/common-buy/CC_content_plugin
2. Click into the model you want to use web search plugin with:
3. Create an app associated with the model:
4. Choose Zero-code mode and One-to-one Chat mode:
5. Configure your bot as follows:
6. Click "Publish" on the upper right corner:
7. Copy bot's endpoint ID and fill into the workflow's configuation page:

🔍 Troubleshooting

1. If you encounter the error "Task not found ......", just click "Automation Task" icon and follow the instructions to install it.

NOTE: this project is based on openai-workflow and modified by fanfank, additional features are added, like reasoning content displaying and switching between models, etc.

---

💡特性

• 使用关键词或默认搜索与 DeepSeek 聊天：
• 使用网页搜索模式（Fn↩︎)获取最新信息：
• 使用强力模式(⌘↩︎)聊天：
• 能显示出推理内容，如果有的话。
• 中断的回答会保留在聊天历史中，并附上"[Answer Interrupted]"。

📝 前提条件

alfred-deepseek 可以配置以下两个模型供应商的其中一个：DeepSeek 官方或火山方舟。 当前火山方舟的 DeepSeek 响应速度快很多，而且支持联网搜索能力，而 DeepSeek 官方还没开放联网搜索 API。 如果你想通过火山方舟来使用 DeepSeek，则按以下步骤操作：

1. 前往 https://www.volcengine.com/ ，创建一个账户并登录；
2. 创建一个新的 API 密钥，可以参考页面 https://www.volcengine.com/docs/82379/1399008；
3. 在 https://console.volcengine.com/ark 页面创建 DeepSeek 模型 Endpoints，例如 DeepSeek-V3 和 DeepSeek-R1：
4. 点击进入模型页面并点击"模型推理"按钮，按照指示完成模型激活和 Endpoint 创建：
5. 从以下页面查看模型 Endpoint：https://console.volcengine.com/ark
6. 确保你使用的是 Alfred 5.5 或更高版本。

📦 安装

双击DeepSeek_by_fanfank_xxxx.alfredworkflow文件即可完成安装。

🔧 配置

配置 Workflow，然后就可以开始聊天了。🚀

不要忘记添加 DeepSeek 到默认搜索！

如果想要使用网页搜索模式，需要按照以下步骤进行配置：

1. 通过这个地址开通联网插件：https://console.volcengine.com/common-buy/CC_content_plugin
2. 点击你想要使用联网插件的模型：
3. 创建一个与模型关联的应用：
4. 选择零代码模式和一对一聊天应用：
5. 配置你的 bot，如下：
6. 点击右上角的"发布"按钮：
7. 复制 Bot 的 endpoint ID，并将其填写到配置页面中：

🔍 故障排除

1. 如果无法正常执行，且在 debug 日志中看到 "Task not found ......" 错误，只需点击 "Automation Task" 图标并按照提示安装即可。

NOTE: 本项目基于 openai-workflow 修改，添加了推理内容显示和模型切换等功能。