Kairos and auto-dream are more interesting than anything in the agent loop section. Memory consolidation between sessions is the actual unsolved problem. The rest is just plumbing tbh

Projects like Beads help with memory consolidation by making it somewhat moot, since it stays "offline" and can be recollected at any moment.

Faced this too. Tried https://github.com/rtk-ai/rtk to compress cli output but some commands started failing and the savings were minimal. Ended up just being more deliberate about context size instead of adding more tooling on top

Ended up with jetbrains mono too, which is what i use anyway. Funny how you just end up defending whatever you're already used to

claude.md rules that cut "great question! here's what i'll do..." are fine. Rules that cut the actual thinking steps break the output. Don't confuse the two.

This is really cool. Most dark mode tools break images, so this solves a real problem.

I like that it runs locally too. Starred it

Thanks, really appreciate it. Hearing that it solves a real problem means a lot to me. Running locally is definitely about privacy, but it partly comes from my past working in a factory where the network was unreliable, so I was basically forced to build tools that worked offline. Knowing that habit is appreciated fills me with joy.

That's a great origin story. Constraints make better software.

Good read. I get the point about real world usage, but I still feel like Neo might fall short with how fast things are getting heavier, especially with Node and modern dev workflows.

Feels fine today, but not sure how well it holds up a couple years down the line.

You know A) most people aren’t doing “dev workflows” and B) Apple sells other computers?

Sure. Was talking about my use case. Probably should've been clearer.

One thing to note is that you can apparently get a ~30% heavy use speedup by adding a thermal pad between the CPU and the case. Stock, the Neo has a 6W TDP, but thermal throttles to 3W after <1s of multicore work because there is no heat dissipation for the chip. With an $8 thermal pad, the CPU can stay at 6W because it no longer hits 105C instantly.

good tip, just offload the heavy stuff to a vps and keep the neo for light tasks. thermal pad helps but why fight the hardware when you can throw compute elsewhere

This is really helpful. I always assumed the migration itself would be the hardest part, but the CI point makes a lot of sense.

How has reliability been so far compared to GitHub, especially on bigger repos?

The PLAN.md question is the one worth pulling on. Once the plan lives in git or the PR it's already downstream of intent and whoever defined what to build has already handed off. The harder problem is giving agents access to the original intent, not just the implementation plan derived from it. When there's drift between what was planned and what got built, a git-resident PLAN.md makes it hard to trace back to why the decision was made in the first place.

The plan will always be downstream of intent though. At least in git you can track the evolution of the plan over time and hopefully annotate the rationale for changes in direction.

Fair point. Git helps track how the plan changes, but it doesn’t always capture the original intent behind it.

I’m glad someone is finally bringing this up, thank you!

What are you suggesting instead? To share the prompt in order to capture the intent? Usually I expect the plan to reflect the prompt.

I find it interesting when I create a PR after a quick session: the description really captures the intent instead of focusing on the actual implementation. I think it’s because the context is still intact, and that’s very useful.