POC: Copilot CLI extension to simplify RPI agentic development workflow#16228
POC: Copilot CLI extension to simplify RPI agentic development workflow#16228JennyPng wants to merge 21 commits into
Conversation
| @@ -0,0 +1,100 @@ | |||
| # Phase 2 — Assumptions (read-only) | |||
|
|
|||
| You are in the **assumptions** phase. Surface every assumption I might be making about | |||
There was a problem hiding this comment.
for /rubber-duck could you pull that command into this skill to get it to help identify assumptions or is the assumption portion supposed to be conducted by the same LLM and then you rubber duck later (or separately)?
There was a problem hiding this comment.
so since CLI extension only works within the same Copilot session, I couldn't achieve the full clean context between stages. So this entire flow is orchestrated by an agent in the same session, but the extension can run /compact between stages which I think achieves a similar effect
but I do have a 'rpi-judge' command that will run rubber duck, and a command like rpi-autojudge where if you run that, and then start this flow, then every single stage in this flow will have rubber duck run after the artifact is generated
There was a problem hiding this comment.
edit: i found out you can actually spawn subagents on demand, it's just experimental but i added that as an option too that you can toggle
2 participants
read README.md for how to use. TLDR check out this branch, install the extension, symlink if you want to access it in any repo otherwise it only works in this repo
automates the research->plan->implement workflow with a Copilot CLI extension, allowing you to manage phases of the workflow with slash commands like
rpi-start,rpi-continue, etc. - evenrpi-autoto run through the entire RPI automaticallytoggling
rpi-autojudgeenables each stage to auto-run rubberduck afterwardsrequires copilot CLI to be on experimental mode