Loop Engineering - more autonomous development

Ideas Feature Requests
,
1

I think Cursor is ready to take development to the next level with implementing a Loop Engineering support feature.

image
image397Ɨ651 4.36 KB

They way I envision it working:

  • Project manage agent - he is the one that will tie all agents together and orchistrate.
  • Research agent - he researches online or from best practices how the feature should work - more from user experience point of view.
    He can possibly create the acceptance criteria for the feature (if user did not provide) or expand on the users acceptance criteria
  • Architect Agent - based on the research and the current application he plans out the feature (Planning mode.)
  • Developer Agent(s) - these guys hit the ground
  • Test Agent(s) - for each dev agent we need a testing agent to write tests and run them to give feedback to the Dev agent.
  • Review Agent - he reviews the feature to the acceptance criteria - then either loops back or accepts.

In setting up the Loop Template, the less the user defines the more the Agents are able to expand to improvise based on the Research / Plan steps.

5

I havenā€™t been keeping up with all the new features of Cursor.

But I have seen the terms and features listed below floating around.

I havenā€™t used them myself yet, but I wonder if they are related or could be of any assistance.

01. Subagents https://cursor.com/docs/subagents

02. Cloud Agents https://cursor.com/docs/cloud-agent

03. Automations https://cursor.com/docs/cloud-agent/automations

04. ā€œMultitaskā€ mode

I canā€™t find docs for ā€˜Multitaskā€™ mode (accessible at bottom of prompt UI) on https://cursor.com/docs.

05. And some possibly related articles:

  • https://cursor.com/help/ai-features/multi-agent

-https://cursor.com/blog/self-driving-codebases

Apologies if any or all of the above concepts are not related to your question.

Unfortunately, I donā€™t yet know the relationship or difference between all of the concepts above, and in what scenarios you would use each one.

Maybe someone else here does :slight_smile: .

A Google Search prompt I just tried was:

Based on your understanding of `https://cursor.com/docs/subagents\`, `https://cursor.com/docs/cloud-agent\`, `https://cursor.com/docs/cloud-agent/automations\`, `https://cursor.com/help/ai-features/multi-agent\` and `https://cursor.com/blog/self-driving-codebases\`, what is the relationship or difference between all of the concepts above (Subagents, Cloud Agents, Automations), and in what scenarios you would use each one.

7

Its kind of taking these concepts one step forward (or going a step higher).