I just published my homebrew coding agent orchestrator I’ve been using in the last - checks calendar - three months.

It’s called littlefactory (repo here) and is a cute little cli that smashes together Ralph loops and a lightweight spec driven workflow, but is not too opinionated and the workflow can be ignored as long as you get to write the implementation tasks in a properly formatted tasks.json.

I mostly use the workflow this way:

1. I start with /lf-explore to rubber duck a change
2. Once happy with the exploration I do /lf-formalize to generate a plan and some specs
3. I then do /lf-do which just runs littlefactory against the change’s todo list. This spins up one new sub agent per task, in sequence, until we run out of tasks.
4. I wait for the implementation phase to finish, I am out of the loop, the main chat interface is just orchestrating so I can just start exploring another change or do something else.
5. Once the implementation is done I /lf-verify to validate it is meeting the specs

That’s it. For peace of mind I configured it to run the sub agents in my sandbox thing.

After that I might tweak the small things and run some review workflows.

Check it out.