mcpField
M
Mcp-ChatBI

Mcp-ChatBI

github.com/dynamicluChat Apps51Updated 75w ago
Repository

About Mcp-ChatBI

Mcp-ChatBI is a ChatBI data analysis agent solution built on the MCP architecture. It fetches figures through tools backed by a Headless BI service or API rather than letting the model generate queries, aiming to keep data accurate.

Key features

  1. 01Retrieves data through tools instead of model-generated SQL
  2. 02Splits multi-metric questions and calls tools in sequence
  3. 03Generates analysis chains automatically and runs them step by step
  4. 04Summarizes results after tool execution
  5. 05Offers API and Gradio startup modes

Use cases

  1. 01Querying several business metrics in one conversation
  2. 02Running comparison, drill-down or attribution analysis
  3. 03Querying BI data through natural language chat

MCP support

Mcp-ChatBI acts as an MCP client that calls tools provided by a Headless BI service or API, with the LLM planning the call order.

Details

From the project README

Mcp-ChatBI

介绍

基于 Mcp 架构的 ChatBI，是一种数据分析智能体的解决方案。

本项目主要解决ChatBI常见的三个问题

  • 1、如何保障数据的100%的准确性？

    由于模型存在幻觉，无论是NL2SQL、NL2Code，都无法保障数据100%的准确。且数据的准确性又是BI系统的红线，因此，本项目使用NL2Tools的方案，Tools可以是Headless BI的服务，也可是API。

  • 2、一次对话如何查询多个指标？

    过去一次对话只能查一个指标，若多个指标需要工程层面去拆解，架构的复杂度非常高，本项目，利用模型的任务规划、推理能力，模型自动拆解多个指标，并按照顺序调用Tools，并返回结果。

  • 3、如何让数据分析的链路自动化？

    常见的数据分析方法，如对比分析、多维钻取、归因运算等，它常常伴有复杂的逻辑推理，当前大模型能力突飞猛进，已经具有复杂问题的推理能力，因此，本项目，利用模型推理能力，自动生成数据链路，并逐次调用Tools，返回结果，最后总结分析。

部署

1. 环境配置
  • 确保你的机器安装了 Python 3.10 - 3.12
# 拉取仓库
$ git clone https://github.com/dynamiclu/Mcp-ChatBI.git

# 进入目录
$ cd Mcp-ChatBI

# 安装全部依赖
$ pip3 install -r requirements.txt
  • 大模型配置
$ vim config/config.toml
[model]
qwen_api_key = "sk-**********"
qwen_model_name = "qwen-max"
2. 启动接口
# 启动API
$ python3 main-api.py
3. 启动Gradio
# 启动Gradio
$ python3 main-webui.py
4. 演示

https://www.bilibili.com/video/BV1b95vzPEjf/

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.