Egregoros · Phoenix Framework

Timeline

Post

Remote status

Context

Matt Hamilton

@eriner@noauthority.social remote

If you're using an LLM for programming, please describe your workflow.

I use https://pi.dev/ as my input.

I feed it a paragraph description of what I want + a gherkin spec (https://cucumber.io/docs/gherkin/reference/) then have it ask me questions and propose changes/additions to the markdown + gherkin spec for this (t)ask.

Then I spawn a new (fresh context) pi to read the enhanced spec and create code in a feature branch.

This is the best workflow I've used so far. Still using Qwen3.6-35B-A3B

Replying to @eriner@noauthority.social

Matt Hamilton

@eriner@noauthority.social remote

I spend most of the time writing the Gherkin spec. The written description is usually only used to provide background information to nudge/inform it of other relevant code/features/info that it should read before spec enhancement.

Replies

Replying to @eriner@noauthority.social

Chromeratt

@chromeratt@noauthority.social remote

@eriner I use GitHub speckit primarily. Experimenting with open spec, in Claude (personal) copilot (work)

"Spec Driven Development" sounds like big upfront design, but with these frameworks it's not. It's small changes, 2 to 5 use cases.

You give the agent as little or much detail as you want. It's a multi-step generative process:

Spec (requirements, uses cases)
Research ("decisions")
Plan (phased, what code to write)
Tasks: fine grained steps