01: Conversation History

A one-shot chatbot forgets unless we pass history back in.

from __future__ import annotations

from pathlib import Path

from agent_patterns_lab.runtime import Message, MockLLM, Tracer


def main() -> None:
    tracer = Tracer()
    model = MockLLM(
        [
            "Got it: you like tea, local food, and easy walking.",
            "I still remember your preferences: tea, local food, and easy walking.",
        ]
    )

    messages: list[Message] = []

    def chat(user_text: str) -> str:
        messages.append(Message(role="user", content=user_text))
        answer = model.complete(messages, tracer=tracer)
        messages.append(Message(role="assistant", content=answer))
        return answer

    print(chat("Remember this: I like tea, local food, and easy walking."))
    print(chat("What preferences did I just give you?"))

    trace_path = tracer.export_jsonl(Path(".traces") / "01_conversation_history.jsonl")
    print(f"[trace] {trace_path}")


if __name__ == "__main__":
    main()

Run:

uv run python examples/01_conversation_history.py

The important change is that each user and assistant message is appended to messages. The next call sees the conversation so far.

This helps the travel assistant remember preferences, but it still returns free-form text.

Next: 02: Structured Output.