Pi is without contest the best harness. Because it’s yours, the total control is the only way to move from driving agents to orchestrating them.

My interruptions and frustration with llms has gone down a cliff. It’s amazing to see what just a few lines of system prompt really can do to keep agents doing the correct thing much more often.

▲

neop1x 18 minutes ago | parent | next [-]

Exactlty. I love that it works the way I want it to work. I can install and write extensions I want.

▲

regexorcist 6 hours ago | parent | prev | next [-]

> Pi is without contest the best harness. Because it’s yours

You can do this with any harness. I really dislike the Node.js ecosystem and avoid Pi and the like. I started with another one written in Go that has all the basics and I can extend it in minutes for anything I need while keeping track of upstream.

▲

brabel 5 hours ago | parent [-]

Pi can run on Bun. I use it within ThinkRail which is an electronBun app! Pretty good speed and low enough memory usage. Highly recommended.

▲

slopinthebag 5 hours ago | parent [-]

bun is just an alternative runner for the node ecosystem

▲

brabel 5 hours ago | parent [-]

I know but ElectronBun installs as a normal binary , you don’t need to install Node or Bun.

	▲	slopinthebag 4 hours ago \| parent [-]
		...it just bundles bun + the source files, it's no different from running a global bun binary.

▲

CarlitosHighway 5 hours ago | parent | prev | next [-]

Is OpenCode already passé? Especially with OpenChamber on top I don't see a reason to switch to PI.

▲

wannabe44 an hour ago | parent | next [-]

Their security track record.

https://wren.wtf/shower-thoughts/stop-using-opencode/

▲

0xbadcafebee 4 hours ago | parent | prev [-]

OpenCode is a worse design in several ways. But the worst part is they don't really respond to issues or pull requests and support is nonexistent. Been in multiple issues with a dozen people begging for a fix with no response. So we all do what we can but there's no one community fix. A ton of OpenCode functionality gets broken constantly because they have almost nonexistent regression tests nor e2e tests for all the features. When they do break it you have to go to Discord and hope someone from their team sees that they broke something, and hope they want to fix it or take a PR for it. This is also ignoring the massive memory leaks that were just accepted as normal.

But with Pi being barebones and intended to be customized, there are a million extensions to choose from compared to OpenCode. This means not only can you write your own extension easily, 5 other people already have. I forked one popular question-asking tool and added support for an external ui (Zenity). And an LLM provider (Synthetic) had a bug which I helped get fixed in a 3rd party extension for the provider. Both I got done in days, because I didn't have to go through some official Pi team; extensions distribute the work better. It's still hard to manage all the extensions, but better than the alternative.

	▲	eddythompson80 4 hours ago \| parent [-]
		Why would you not ask opencode to fix the issue for you instead of asking opencode folks to ask their opencode to fix it for you? Surely Maintaining your own fork is just one markdown file and one corn task away.

▲

julesrms 3 hours ago | parent | prev | next [-]

> it’s yours, the total control is the only way to move from driving agents to orchestrating them

I wrote Juggler (https://github.com/juggler-ai/juggler) because I liked Pi's approach to everything being customisable and controllable, but I wanted a slick, navigable GUI rather than terminal hell.

...BUT: having gone to great pains to build an architecture around customisation, where every possible pluggable thing is a plugin, I'm starting to feel jaded on the whole "we need to customise it all" mentality. The models have got so good over the last year that I just don't think that tweaking the harness is something many people will need to do as things progress. The models only need a bare-bones set of tools to do everything they need, and I'm starting to think that UX and UI are the only really important things to focus on in an agent.

(Although from the perspective of coding a harness, making things plugins and having a plugin API does help to keep things clean and stops the codebase getting sloppy)

▲

Rodmine 5 hours ago | parent | prev | next [-]

>Pi is without contest the best harness.

There is contest.

https://frontierharness.org/

With minimal work, you can get DSH web to be exposed to your internet behind Cloudflare Access (or similar) and have a harness running on your computer at home from any browser, including your phone's.

	▲	novaRom 2 minutes ago \| parent \| next [-]
		It's outdated site, e.g. they tested Pi version 0.84.2 while Pi 1.0 is available since October 1.
	▲	nairboon 2 hours ago \| parent \| prev [-]
		The cost leader of this benchmark, "Exo Harness", seems to a project archived already, or what harness is this?

▲

tonfreed 7 hours ago | parent | prev [-]

Second this, I went from Claude code fighting me all the time and being insanely opaque with its internals to Pi with detailed telemetry about what my agents are actually spending time on. My entire harness is completely custom, and works exactly the way I want it to. Absolutely love Pi