This is part 2 of a series on building a fully functional terminal agent from scratch.

You can find the part 1 of this series in the following page: https://dimiro1.dev/building-an-agent-from-scratch

The complete source code is available in my GitHub: https://github.com/dimiro1/agent-from-scratch/tree/main/02

In this second part, we'll implement the conversation loop so we can have real back-and-forth conversations with the LLM. The agent will remember our past messages from the same conversation, making interactions feel natural and contextual.