connectai-e/talkbase

💬 Write data into lark-base tables in a chat way，power by TypeChat.

Introduction · Deploy Your Own · Setting Up Locally · Tech Stack · Contributing · License

English · 简体中文

Introduction

TalkBase is an open-source AI chatbot that uses TypeChat and the Vercel AI SDK to interact with natural language.

🐵 Easily enter data by natural language , with the requirements table field meta and descriptions.

🐵 Only English？ no! any langguage such as Chinese

🙈 automatically describe the purpose of the table and give text suggestions for entering data

🙈 perfectly compatible with field annotations of multi-dimensional tables, add more field information

Quick Start

1️⃣ open the grayscale qualification for base extension

https://applink.feishu.cn/client/chat/chatter/add_by_link?link_token=c55n4142-fce8-4792-b851-f92c9c7d8300

2️⃣ copy talk-base online link to the extension script address

https://talkbase.connectai.site/

3️⃣ start to chat-base ~

Dev Local

To set up talkBase locally, you'll need to clone the repository and set up the following environment variables:

• OPENAI_API_KEY – your OpenAI API key (you can get one here)

1️⃣ clone deme

git clone https://github.com/ConnectAI-E/BaseScript-LinkPreview
pnpm install
pnpm dev

2️⃣ copy talkbase dev link to the extension script address

http://localhost:5173/

Deploy your own

You can deploy your own version of TalkBase with one-click:

Tech Stack

ChatH is built on the following stack:

• Next.js –framework
• TypeChat –llm framework
• BaseScript -lark base script plugin
• Vercel - deployments
• TailwindCSS – css styles

Communication

License

Licensed under the MIT license.

talkbase's Issues

希望可以兼容更多的多维表格字段 Hope to be compatible with more multidimensional table fields

• 多行文本 content
• 超链接 Hyperlink
• 单选 single choice (need define option in advance)
• 多选 multiple choice (need define option in advance)
• 人员 people
• 日期 date
• 数字 number
• 评分 score
• 进度 precent
• 货币 money
• 复选框 radio
• 电话号码 phone number
• 地理位置 position
• 单向关联
• 双向关联

多维表格参考开发资料
https://bytedance.feishu.cn/docx/HjCEd1sPzoVnxIxF3LrcKnepnUf

期待中英的多语言 i18n

Looking forward to distinguishing between Chinese and English multilingual i18n

希望支持配置openai.api_type、api_base 和model_name or engine

使用Azure Openai的话，没办法只配置key，还需要手动指定以上参数才可以链接

2121

@Leizhenpeng

测试issue

作为用户，期待结合表头给一下比较切合场景的示例输入

例如：

根据下面的类型文件结构，判断出此表格的作用后，假设出恰当情景，用人类的自然语言给一个示例输入以确保可以解析出大部分字段数据，不超过20字。
Q： export interface food {
size: "大份" | "中份" | "小份";
meat: string;
staple_food: "米饭" | "面条" | "饺子";
full_content: string;
spicy: "不要辣" | "微辣";
}
A：大份牛肉粒炒饭，加份辣子

Q : export interface hobby {
raw_input: string;
fruits: ("橘子" | "苹果" | "菠萝")[];
}
A: 我一般喜欢吃点橘子和苹果

Q:  export interface SentimentResponse {
content: string;
sentiment: "negative" | "neutral" | "positive";
}
A : 

期望可以通过对话的方式操作多维表格视图

例如：想要输入一段话，可以自动过滤或者排序，操作视图

期待支持多维表格的字段描述

检测到table切换的时候，可以自动读取最新表格的结构

• 通过selection变化检测table now
• 切换table有toast提示
• 检测当前的table，并根据table 的结构猜测当前表格作用，做出文案引导