Inspiration
I wanted to learn about MCP as I did not know anything about it coming into this hackathon. I also wanted to learn more about working with Q and the Q CLI to get a better feel for pair programming with this tool.
What it does
The simple idea with which to potentially exercise my learning requirements was to use the AWS infrastructure to collect asset prices over time and to be able to ask questions about the data using regular English.
How I built it
I acted as the 'idea guy' and architect of the project and had Q write all of the code for me.
Challenges we ran into
Q sometimes has some 'strange' interpretations of what I want implemented and so communication can get patchy at times. Q also has a tendency to hallucinate on implementations especially with regard to MCP server implementations. I could not rely on Q to implement the MCP server correctly at all. Q also has a tendency to revert some minor code edits I made, which it quickly reversed when I yelled at it.
Accomplishments that we're proud of
I met my learning objectives.
I think it is really crazy that I can type an English question about my data and get an on-point English response and lucid analysis of the same data.
What's next for QGold
I can see the project evolving to become a more mature, full-blown financial modeling and analysis tool.
Built With
- amazon-web-services
- boto3
- dynamodb
- eventbridge
- lambda
- python
- q
- sam

Log in or sign up for Devpost to join the conversation.