VecMed-MCP

Instruction

These files are aimed to establish a Milvus Vector Database

What are those file used for?

0. docker-compose.yml: used for initiate MilvusDB
1. setup_milvus.py : formulate the schame of database.
2. download_pubmed_tomilvusdb.py: used for timely update, currently days=30.
3. search_milvusdb.py: only a test of searching function in MilvusDB.
4. llm_process_search_result.py: a experiment of trying to summarize search result using a LLM.
5. updating_milvusdb.log: record the updating operation.

Current ATTU WebUI

See all the data on web, manage the collection, schame. No password needed. Currently, the collection "pubmed_rare_disease_db" has more than 160k data from Pebmed related to rare disease.

Environment Configuration

Launch the docker-compose.yml

docker compose up -d

MCP Tools Preparation

pip install marshmallow==3.20.1 -i https://pypi.tuna.tsinghua.edu.cn/simple\

pip install Flask -i https://pypi.tuna.tsinghua.edu.cn/simple\

pip install pymilvus -U -i https://pypi.tuna.tsinghua.edu.cn/simple\

pip install "mcp[cli]" -i https://pypi.tuna.tsinghua.edu.cn/simple\

Attu WebUI Docker Setup

docker pull zilliz/attu:v2.4.4 docker run -d --name attu -p 8000:3000 -e MILVUS_URL=192.168.10.199:19530 zilliz/attu:v2.4.4

Work Procedure

Notice: you can build whatever database you want, not necessary a medical artical database like I did, the procedure will be most likely.

1. Build your own Milvus database with designed collection schema with the code in the main folder, following the steps 1-5.
2. Download as much data as you need and save them to your vector database, code and instructions in the subfolder - download_pubmed_2015-2025.
3. Luanch the mcp-server with a http api service, or change it to stdin transport as you need, code and instructions in the subfolder - pubmed-mcp-server.
4. Set a timer update the database with download_pubmed_to_milvusdb_2.py
5. Add your mcp-server to the agent/LLM/workflow, here I intergrate it in Dify workflow as an example:

Timer Update the Database

To add a new timer operation: crontab -e To view to timer: crontab -l

Author

David Qu
Undergraduate Researcher | AI Algorithm Engineer
University of Toronto Scarborough - Department of Computer Science
📧 davidsz.qu@mail.utoronto.ca