Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Pi's Minimalism Is Its Advantage

earendil.com

Post reply on HN
Print view View original

111–120 of 307 posts

Re: Pi's Minimalism Is Its Advantage

#111

Post by paldepind2 »

timwis wrote: post #90
swingboy wrote: post #27

Aside from the minimal system prompt, how does it handle context better than other agents? It still has to send the system prompt (which includes AGENTS.md and skill definitions) along with the full conversation every request, no?

The /tree feature is incredible for context management. It's really surprising the other harnesses haven't slurped it up yet. It lets you rewind back to any previous message and fork the conversation from there, removing your 'side quest' (e.g. where you dig into something the agent said) from the context. Some other harnesses have a 'rewind' feature, but this lets you maintain the previous conversation history in a…

You can do that in Copilot in VSCode. There’s a fork icon before every message that creates a new session from that point. The relationships are not maintained in a tree though (not sure if Pi does that).

Re: Pi's Minimalism Is Its Advantage

#112

Post by colordrops »

pavo-etc wrote: post #10

I've had a lot of success running Pi on my server in headless mode and wrapping it in an XMPP client. This means I can talk to it wherever I can access XMPP (everywhere). It also mean agents can talk to each other when they need to. They've got a shared wiki they interact with and github issues as their todo list. I am running several named pi instances in parallel in their own user account on NixOS, so they can inst…

This is cool, I'll have to try this.

Want I've been doing for transit is using termux on android with magisk for root access, allowing me to install Nix home manager and harnesses on my phone, and then drive everything from my phone, including eternal terminal sessions to my homes erver, where it spins up agents there as well.

Incidentally it's also great for debugging issues with my phone.

Re: Pi's Minimalism Is Its Advantage

#113

Post by whazor »

Both Codex and Claude have trained their models to their harnesses.

There are some experimentations by Igor Warzocha to extract Codex shapes and put it in Pi: https://github.com/IgorWarzocha/howaboua-pi-stuff/tree/main/...

I'm expecting every model will have a fine tuned Pi extension at some point.

Re: Pi's Minimalism Is Its Advantage

#114

Post by alexhans »

grewil2 wrote: post #96

Are all of you pi users paying Anthropic and OpenAI for API usage? Or can you combine pi with a subscription? So far API usage is a lot more expensive than subscription and if you need better models than Deepseek and Kimi, and you are not wealthy, I don’t see a way around this.

You can also use PI with local models like Qwen3.5-35B-A3B [1] and they can be surprisingly good if you development in minimalistic/simple ways.

One game changer when it comes to tweaking configs that are optimized for your use case is that you can easily use a "more powerful" cloud model to identify a good enough config for your local server/pi settings combination [2] in a pattern that applies pretty much anywhere.

- [1] https://huggingface.co/Qwen/Qwen3.5-35B-A3B

- [2] https://alexhans.github.io/posts/find-the-loop-story-first.h...

Re: Pi's Minimalism Is Its Advantage

#116

Post by theshrike79 »

hakunin wrote: post #17
suprjami wrote: post #16

Earlier quoted context omitted.

I have the same opinion as your first paragraph, but I don't want to spend weeks or months vibe-coding basic features which come built into almost every other agent. Yeah maybe Claude/OpenCode/KiloCode/Hermes/whatever are not as minimal as Pi but they also work right now.

We probably differ a lot in what we consider "basic". Are subagents basic? I found them only useful in very few situations. Is LSP support basic? There are mixed results on whether it helps or hurts. Are multiple choice asking tools basic? I found that they add extra unnecessary ceremony, eat extra context, and I almost never answer with one of the choices exactly. And if you try vanilla Pi, you will also find out th…

Subagents are good when the harness (And agent?) understand how smart they are.

My root level CLAUDE.md has pretty much just "use a lower tier agent when relevant".

Then I daily-drive Opus, it automatically offloads simpler stuff to Sonnet or even Haiku based its own reasoning because it "knows" their capabilities.

It's so much more cost/token efficient to do it like this. Opus writes the exact implementation plan for Sonnet and then waits for it to complete. After that it checks the work and fixes any issues itself.

In Codex, for example, this doesn't work because the whole system doesn't know about agent tiers and barely can use subagents. So I'm just running Sol all the time.

Re: Pi's Minimalism Is Its Advantage

#117

Post by tudelo »

prettyblocks wrote: post #41
langs wrote: post #38

Pi’s minimalism reveals a simple truth: the LLM is the core of any agent harness. Consequently, much of current harness tuning will become redundant(or even a hindrance) with next-gen models.

I always thought that the frontier coding models were specifically trained to perform well in the harness from their providers. I imagine this will always be the case to some extent.

From what I understand pre-training is totally irrelevant to this and as far as post training goes there will be multiple steps, for claude and codex and the like that ship with a harness, the harness is definitely included in evaluation. However, they will definitely include evaluation from a variety or even none, and settle on something that works the "best" for a release.

Disclaimer: No first hand knowledge

Re: Pi's Minimalism Is Its Advantage

#120

Post by rcarmo »

whazor wrote: post #113

Both Codex and Claude have trained their models to their harnesses. There are some experimentations by Igor Warzocha to extract Codex shapes and put it in Pi: https://github.com/IgorWarzocha/howaboua-pi-stuff/tree/main/... I'm expecting every model will have a fine tuned Pi extension at some point.

I’ve been using both families of models inside pi/piclaw (https://rcarmo.github.io/projects/piclaw/) and I assure you they work _better_ in that environment than in the originals. The models are not trained to the harnesses, the harnesses provide cues that the models follow.
Post reply on HN