What is the best LLM-assisted (like Cursor) coding setup for crewai?

[a-ghorbani]

I've been using Cursor as my primary coding environment and Claude (currently 3.7 Sonnet) as my go-to LLM. this works fairly well in most of the coding tasks, but it almost always gets things wrong when it comes to::

• understanding CrewAI's principles, ie how to properly structure agents, tasks, and processes.
• using CrewAI APIs correctly and often suggesting outdated or completely incorrect function calls, imports etc
  I end up spending a lot of time debugging and manually correcting the output.

I know this space is evolving quickly, and LLMs are trained on outdated docs or codes. To improve the responses, I've tried providing links to the docs/codebase, using examples etc. This helps a bit, but I still have to do most of the heavy lifting myself.

So I’m curious has anyone found a solid workflow (ideally in Cursor) for working with CrewAI?
Would love to hear what’s been working (or not working) for you!