---
title: "Comment by Adam Lewis on \"Designing with a Co-Pilot: Autonomy, Control, and What’s New in Deep Cuts\" | Hashnode"
description: "The skills/ directory pattern is the same in our setup. We've got one markdown file per architectural decision the agent kept getting wrong, and they all get re"
url: "https://hashnode.com/posts/designing-with-a-co-pilot-autonomy-control-and-what-s-new-in-deep-cuts/6a1fb517c5c3ae375fba6ac8/comment/6a25a6e0f416d1b56e8f0068"
---

[Discussion](https://hashnode.com/posts/designing-with-a-co-pilot-autonomy-control-and-what-s-new-in-deep-cuts/6a1fb517c5c3ae375fba6ac8)

Comment [Designing with a Co-Pilot: Autonomy, Control, and What’s New in Deep Cuts](https://hashnode.com/posts/designing-with-a-co-pilot-autonomy-control-and-what-s-new-in-deep-cuts/6a1fb517c5c3ae375fba6ac8)

[Adam Lewis](https://hashnode.com/@nark3d)

[Jun 7](https://hashnode.com/posts/designing-with-a-co-pilot-autonomy-control-and-what-s-new-in-deep-cuts/6a1fb517c5c3ae375fba6ac8/comment/6a25a6e0f416d1b56e8f0068)

The skills/ directory pattern is the same in our setup. We've got one markdown file per architectural decision the agent kept getting wrong, and they all get re-read at the start of every session. After a few sessions there's not much left to re-explain.

Scalability is harder to catch with a hook alone. We've started writing the quantified threshold into the spec ('handles 30k rows in under 2 seconds'), so the agent flags it itself before I'd otherwise need to interrupt.

[Roberto Lupi](https://hashnode.com/@rlupi)

[Jun 8](https://hashnode.com/posts/designing-with-a-co-pilot-autonomy-control-and-what-s-new-in-deep-cuts/6a1fb517c5c3ae375fba6ac8/comment/6a25a6e0f416d1b56e8f0068#reply-6a267cbfdb7c7d5784b1054f)

Yesterday, I have actually make them design and implement a coordination system that let them reach consensus, and we prototyped it already.

See the repo docs:

doc/collab/sessions/2026-06-08-multi-agent-collaboration-research doc/proposals/ccrep-synthesis.md

You can get the source code for the MCP from the [https://github.com/robertolupi/deep-cuts](https://github.com/robertolupi/deep-cuts) repo, but I am going to publish it separately soon.

[Adam Lewis](https://hashnode.com/@nark3d)

[Jun 9](https://hashnode.com/posts/designing-with-a-co-pilot-autonomy-control-and-what-s-new-in-deep-cuts/6a1fb517c5c3ae375fba6ac8/comment/6a25a6e0f416d1b56e8f0068#reply-6a27e0148d068a60ae4d0195)

Nice. The thing I'd point to in what you did is that the consensus ends up written in a doc the agents re-read, rather than only living in the session that produced it. That's the same reason the per-decision markdown files have stuck for me - the agreement survives the context window, so the next run picks it up instead of going over it again. I'll read through the synthesis proposal. Multi-agent coordination is one I haven't spent much time on yet.
