Resume agent mode after long running processes

Feature Requests
1

Feature request for product/service

Chat

Describe the request

I am working on a variety of scientific data pipelines and would like the ability to customize how agent resumes after a process it invokes. My development loop looks something like:

  1. write a long prompt that details the implementation plan
  2. let cursor write code and invoke the pipeline locally (via a custom cli)
  3. let cursor inspect the output in structured logs (grep, jq, awk) and inspect the data (duckdb)

The challenge I have at the moment is being able to configure long running tools as this often times out if we make this blocking. Instead we have a run launch command and a run status command, and we let cursor poll the status until it is ready to analyse (with optional timeouts for the run).

This doesn’t seem particularly efficient and I would rather be able to invoke a call back to tell cursor to “resume agent” on the finishing of the command. Alternatively this could look like background commands (in which case I can just design a sync api instead).