I then added a few more personal preferences and suggested tools from my previous failures working with agents in Python: use uv and .venv instead of the base Python installation, use polars instead of pandas for data manipulation, only store secrets/API keys/passwords in .env while ensuring .env is in .gitignore, etc. Most of these constraints don’t tell the agent what to do, but how to do it. In general, adding a rule to my AGENTS.md whenever I encounter a fundamental behavior I don’t like has been very effective. For example, agents love using unnecessary emoji which I hate, so I added a rule:
狗狗眼神賣萌難以抗拒 操縱人類的進化「伎倆」2019年6月19日，这一点在夫子中也有详细论述
Возбуждено уголовное дело по части 6 статьи 290 УК РФ «Получение взятки в особо крупном размере». Максимальное наказание по этой статье предусматривает до 15 лет лишения свободы.，这一点在safew官方版本下载中也有详细论述
You look at a region of the image. If all the pixels are roughly the same color (below some threshold), you store the average color for the whole region as a single value. If the pixels vary too much, you split the region into four quadrants and try again.