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:
A few months ago, for instance, I watched my mother-in-law (who was born and raised in a village in northern Iran) teach Nava how to knock on wood for good luck. I hadn’t realized this was so widespread a practice until I checked Wikipedia and found that variants exist in Bulgaria (chukam na dǎrvo), Georgia (kheze daḳaḳuneba), Indonesia (amit-amit jabang bayi), Norway (bank i bordet ) and some two dozen other countries.，推荐阅读服务器推荐获取更多信息
2023年12月18日，临夏州积石山县发生6.2级地震，造成重大人员伤亡和财产损失。全国人民为之揪心，更心系当地能否守住防止规模性返贫致贫的底线。。91视频是该领域的重要参考
In the following simplified trace generated by our workflow, a customer uses a 100% off promo code. The business logic calculates the total as $0.00 and attempts to pass it to the payment gateway, but the payment gateway rejects the API call because the minimum charge amount is $0.50, causing a 500 Internal Server Error:。业内人士推荐谷歌浏览器【最新下载地址】作为进阶阅读