I’m a heavy Cursor user.
At one point, high usage actually made me lose confidence in the platform it became hard to work efficiently.
But after seeing the direction Cursor is taking, especially with Auto Mode + Composer, my trust came back.
For me, this combination is a real win.
Since then, I managed to build an application from start to finish clean code, clear structure, and solid architecture.
Plan Mode is good, but not perfect, so I use it differently:
My workflow:
I use Plan Mode in Cursor to structure the feature
Then I take that plan to GPT-5.2
I give GPT full project context: docs, structure, architecture
I let Cursor and GPT “talk” reviewing decisions, suggesting improvements, checking libraries, versions, and architecture
After that, I document the improvements and execute them back in Cursor
This loop helped me build something I’m genuinely proud of.
Honestly, it feels like nothing can hold me back anymore.
My current setup:
Cursor pro
Claude Code cli
GPT subscription
That’s it. I don’t feel I need anything else right now.
I also work with MCP integrations (Railway, RunPod, Supabase,github.etc), which makes a big differe Cursor can see everything, and I can ask better questions, verify decisions, and double-check logic.
I always start a new chat per session to keep context clean.
Use cursor rules and update them .
I’d love to hear:
How do you use Plan Mode?
Any tips to improve this workflow?
Anything you’d change or optimize?