Prompt-It

Inspiration

We wanted to make customizable agents accessible to people with little to no tech knowledge. Currently setup frameworks like OpenClaw only cater to developers or people with more skills in technology, we want to limit the skill gate necessary to customize agents.

What it does

We provide a easy to use user interface that does not require the user to install any frameworks or set up. The user is able to customize their own agents through selecting nodes and paths on a "skill tree" that makes the set up easy and fun.

How we built it

We used React to design the frontend webpage with a Python based backend, which FastAPI is used to bridge communication between the two. We implemented a custom agent manager enables controlled multi-session execution, NLP based text recognition model to adjust model skills based on system prompt, and a JSON skill-tree system makes agent behavior modular and extensible.

Challenges we ran into

One main challenge we had was optimizing the response time of the customized agents and the run time of our program.

Accomplishments that we're proud of

We are most proud of implementation of the feature to run multiple different agents with session memory stored. Allowing for users to switch between different agent set ups through changing session ID.

What we learned

We learned how to efficiently prompt agents through command line interface and integrate natural language processing models into our program to detect specific user prompts for our agent set up.