Feature request for product/service
Cursor IDE
Describe the request
I often work in Cursor with an iterative planning workflow:
- I draft a plan with a strong model (e.g., “GPT High”) and model everything writes in “.md” document. (Chat #1)
- I send that document to another model (e.g., “Opus”) to critique it (missing cases, risks, edge cases, alternative approaches). (Chat #2)
- I bring the critique back to the first chat and ask it to evaluate whether the critique is reasonable, then I refine the plan. (return to Chat #1)
I repeat the same iteration again (sometimes with a different reviewer model).
This produces noticeably better plans, but it’s repetitive: the prompts are very similar each time and the pipeline is basically identical.
P.S. Of course I use commands etc…
Request
Please add a way to automate a “plan / do smth in chat 1 with selected model X → review in chat 2 with selected model Y → return document / result to chat 1” pipeline inside Cursor.
It’s like pipeline builder.
What I’d like to see
Multi-step agent/workflow builder (like a small “pipeline” editor):
Why this matters
Saves time and reduces copy/paste between chats.
Increases plan quality by systematically catching edge cases.
Makes “best practice” planning repeatable across tasks and team members.
If helpful, I can provide example prompts/rubrics I currently reuse for critique and adjudication.
I think these pipelines can be brand new feature in Cursor and can automate a lot of stuff like: code review processes, bug finding, separate huge features in different steps and lot of different stuff. It will be helpful to run in 1 button chain of actions with different model. That brings more strong and robust results.
Thanks in advance.