Thanks @Tom Welsh I will look at the github repo.

I also find this repo: https://github.com/PatrickJS/awesome-cursorrules

Here’s a clearer and more polished version of your message: Hi Paul, Thank you for your reply. I’ve been working on writing a unit test for a Game application. To generate the unit test, I used GPT-4 and LangChain. I should mention that I did not use an agentic approach. This was my first project using an LLM. The main issue I faced was that the code relied heavily on complex objects. Without knowing the definitions of these objects, it was challenging for the LLM to accurately mock them during testing. To address this, I had to pass a large amount of context—essentially providing several Python files—to ensure the LLM had the necessary object definitions for mocking. In many cases, the LLM failed to mock the objects correctly, and in some cases, I ran out of context because the function I planned to test depended on too many external objects. Here’s the general approach I followed: 1. Determine if the function is testable using the LLM. 2. Extract all dependencies required for testing the function. 3. Include the code for all external dependencies in the LLM prompt. 4. Write test scenarios based on the function behavior. 5. Generate unit tests from the scenarios. Despite these steps, I encountered challenges when working with functions requiring many complex dependencies. Below is an example of the code I attempted to write a unit test for, along with the source code. *---------------------------------------------< source code >----------------------------------------------------* from dataclasses import dataclass from typing import List from dataclasses_json import dataclass_json from domain import BotAction, BoardReference, BotActionOutcome from games.common.actions import DamageAttackOutcome, BotwarAction @dataclass_json @dataclass class RangeAttackOutcome(DamageAttackOutcome): energy_spent: int class RangeAttackBotAction(BotwarAction): @staticmethod def _do_range_attack(board: BoardReference, attacks: List[BotAction]) -> List[BotActionOutcome]:

@Paul Miller Thanks for sharing.