Counselling Agent

Capabilities:

1. RAG using Pandas on a dataset from Amod's mental helath counselling conversations dataset
2. Google Gemini 2.0 Flash model in use.
3. Use of keyword extraction/inference from the conversation to ease advice development.

How to Use:

1. Install Prerequisites

• In the repo folder, run pip install -r requirements.txt to install the required packages. (preferably in a virtual environment if you'd like to keep the dependencies isolated)
• For the developers: if you'd like to run tests, do pip install -r requirements-dev.txt.

2. Running the Agent:

There are two ways:

1. Run python ./agent/global_agent.py in the root (or run that file within its parent folder; either will work)
2. Create a new Python file in the root, add the following:

from agent import *

if __name__ == "__main__":
    agent = AIAgent()
    agent.run()

Possibly Unwanted Pieces of Code:

1. Function write_cache in ./agent/global_agent.py: I don't really think we'll need to cache formerly accessed dataset rows, it's extra memory consumption in the cloud, if ever done so.

Todo:

• Add MCP interface to play soothing music if required in the chat interface.