19 comments

  • rcarmo1 hour ago
    My current fave harness. I&#x27;ve been using it to great effect, since it is self-extensible, and added support for it to <a href="https:&#x2F;&#x2F;github.com&#x2F;rcarmo&#x2F;vibes" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;rcarmo&#x2F;vibes</a> because it is so much faster than ACP.
    • badlogic1 hour ago
      wow, i love this! was about to build this myself, but this looks exactly what i want.
      • rcarmo56 minutes ago
        The better web UI is now part of <a href="https:&#x2F;&#x2F;github.com&#x2F;rcarmo&#x2F;piclaw" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;rcarmo&#x2F;piclaw</a> (which is essentially the same, but with more polish and a claw-like memory system). So you can pick if you want TS or Python as the back-end :)
        • badlogic54 minutes ago
          if i ever want a claw, i&#x27;d obv. go with this :)
          • rcarmo37 minutes ago
            The claw version’s web UI essentially has better thinking output, more visibility of tool calls, and slightly better SSE streaming. I’ve backported some of it to vibes, but if you want to borrow UI stuff, the better bits are in piclaw. I use both constantly on my phone&#x2F;desktop.
  • mccoyb38 minutes ago
    Pi has made all the right design choices. Shout out to Mario (and Armin the OG stan) — great taste shows itself.
    • semiinfinitely19 minutes ago
      I do not understand why in the age of ai coding we would implement this in javascript
      • mccoyb14 minutes ago
        It’s straightforward: JavaScript is a dynamic language, which allows code (for instance, code implementing an extension to the harness) to be executed and loaded while the harness is running.<p>This is quite nice — I do think there’s a version of pi’s design choices which could live in a static harness, but fully covering the same capabilities as pi without a dynamic language would be difficult. (You could imagine specifying a programmable UI, etc — various ways to extend the behavior of the system, and you’d like end up with an interpreter in the harness)<p>At least, you’d like to have a way to hot reload code (Elixir &#x2F; Erlang could be interesting)<p>This is my intuition, at least.
      • alvivar13 minutes ago
        One thing in its favor is that it was kind of easy and simple to extend and contribute to and share extensions. JavaScript&#x2F;node is still the biggest dev environment. The same can be said for OpenClaw.
      • Blackarea14 minutes ago
        yes! I just don&#x27;t understand that as well. Up until some time ago claud code&#x27;s preferred install was a npm i, wasn&#x27;t it? Please serious answers for why anyone would use a web language for a terminal app
  • muratsu11 minutes ago
    I’m working with a friend to build an ui around Pi to make it more user friendly for people who prefer to work with a gui (ala conductor). You can check out the repo: <a href="https:&#x2F;&#x2F;github.com&#x2F;philipp-spiess&#x2F;modern" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;philipp-spiess&#x2F;modern</a>
    • ramoz7 minutes ago
      In the same spirit, I also ported a planning UI extension for Pi.<p><a href="https:&#x2F;&#x2F;plannotator.ai&#x2F;blog&#x2F;plannotator-meets-pi&#x2F;" rel="nofollow">https:&#x2F;&#x2F;plannotator.ai&#x2F;blog&#x2F;plannotator-meets-pi&#x2F;</a>
  • rahimnathwani33 minutes ago
    Hugging Face now provides instructions for using local models in Pi:<p><a href="https:&#x2F;&#x2F;x.com&#x2F;victormustar&#x2F;status&#x2F;2026380984866710002" rel="nofollow">https:&#x2F;&#x2F;x.com&#x2F;victormustar&#x2F;status&#x2F;2026380984866710002</a>
  • himata41131 hour ago
    Preconfigured PI: <a href="https:&#x2F;&#x2F;github.com&#x2F;can1357&#x2F;oh-my-pi" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;can1357&#x2F;oh-my-pi</a>
  • mongrelion41 minutes ago
    <i>Pi ships with powerful defaults but skips features like sub-agents and plan mode</i><p>Does anyone have an idea as to why this would be a feature? don&#x27;t you want to have a discussion with your agent to iron out the details before moving onto the implementation (build) phase?<p>In any case, looks cool :)<p>EDIT 1: Formatting EDIT 2: Thanks everyone for your input. I was not aware of the extensibility model that pi had in mind or that you can also iterate your plan on a PLAN.md file. Very interesting approach. I&#x27;ll have a look and give it a go.
    • ramoz39 minutes ago
      See my comment in the thread but there is an intuitive extension architecture that makes integrating these type of things feel native.<p><a href="https:&#x2F;&#x2F;github.com&#x2F;badlogic&#x2F;pi-mono&#x2F;tree&#x2F;main&#x2F;packages&#x2F;coding-agent#extensions" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;badlogic&#x2F;pi-mono&#x2F;tree&#x2F;main&#x2F;packages&#x2F;codin...</a>
    • alvivar38 minutes ago
      I plan all the time. I just tell Pi to create a Plan.md file, and we iterate on it until we are ready to implement.
    • miroljub38 minutes ago
      Check <a href="https:&#x2F;&#x2F;pi.dev&#x2F;packages" rel="nofollow">https:&#x2F;&#x2F;pi.dev&#x2F;packages</a><p>There are already multiple implementations of everything.<p>With a powerful and extensible core, you don&#x27;t need everything prepackaged.
  • suralind26 minutes ago
    I’ve been testing it for a few days on pretty much clean install (no customizations&#x2F;extensions) and it’s ok. Not sure if I like it yet.
  • ramoz42 minutes ago
    The way you’re able to extend the harness through extension&#x2F;hook architecture is really cool.<p>Eg some form of comprehensive planning&#x2F;spec workflow is best modeled as an extension vs natively built in. And the extension still ends up feeling “native” in use
  • jmorgan1 hour ago
    I&#x27;ve been using Pi day to day recently for simple, smaller tasks. It&#x27;s a great harness for use with smaller parameter size models given the system prompt is quite a bit shorter vs Claude or Codex (and it uses a nice small set of tools by default).
    • rpastuszak7 minutes ago
      Which models do you use and what for? I&#x27;m looking for ideas to play with.
  • arjie1 hour ago
    Has anyone used an open coding agent in headless mode? I have a system cobbled together with exceptions going to a centralized system where I can then have each one pulled out and `claude -p`&#x27;d but I&#x27;d rather just integrate an open coding agent into the loop because it&#x27;s less janky and then I&#x27;ll have it try to fix the problem and propose a PR for me to review. If anyone else has used pi.dev or opencode or aider in this mode (completely non-interactive until the PR) I&#x27;d be curious to hear.<p>EDIT: Thank you to both responders. I&#x27;ll just try the two options out then.
    • fred_tandemai43 minutes ago
      Been using pi exactly for this and it&#x27;s working great!
    • evalstate45 minutes ago
      fast-agent lets you do this as well (and has a skill in its default skills repo to help with automation&#x2F;running in container&#x2F;hf job).
    • rcarmo1 hour ago
      You probably want to look into pi then - it&#x27;s extremely extensible.
    • dosinga1 hour ago
      you can run <a href="https:&#x2F;&#x2F;block.github.io&#x2F;goose&#x2F;" rel="nofollow">https:&#x2F;&#x2F;block.github.io&#x2F;goose&#x2F;</a> in headless mode (I work on goose)
  • infruset1 hour ago
    Note there is a fork oh-my-pi: <a href="https:&#x2F;&#x2F;github.com&#x2F;can1357&#x2F;oh-my-pi" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;can1357&#x2F;oh-my-pi</a> of <a href="https:&#x2F;&#x2F;blog.can.ac&#x2F;2026&#x2F;02&#x2F;12&#x2F;the-harness-problem&#x2F;" rel="nofollow">https:&#x2F;&#x2F;blog.can.ac&#x2F;2026&#x2F;02&#x2F;12&#x2F;the-harness-problem&#x2F;</a> fame. I use it as a daily driver but I also love pi.
  • lukasb29 minutes ago
    But I can&#x27;t use my Codex plan with it, right? I have to use an API key?
    • mccoyb28 minutes ago
      You can use your Codex plan with it. OpenAI endorsed it several weeks ago, as far as I remember. That could change, however, but now seems safe.
  • ge961 hour ago
    Is that an official term &quot;coding harness&quot;<p>Wondering if you wanted a similar interface (though a GUI not just CLI) where it&#x27;s not for coding what would you call that?<p>Same idea cycle through models, ask question, drag-drop images, etc...
    • arcanemachiner8 minutes ago
      Yes. It seems to be the term that stands out the most, as terms like &quot;AI coding assistant&quot;, &quot;agentic coding framework&quot;, etc. are too vague to really differentiate these tools.<p>&quot;harness&quot; fits pretty nicely IMO. It can be used as a single word, and it&#x27;s not too semantically overloaded to be useful in this context.
    • rcarmo57 minutes ago
      LLM harness has been in vogue for a year now…
      • outofpaper53 minutes ago
        A harness is a collection of stubs and drivers configured to assist with automation or testing. It&#x27;s a standard term often used in QA as they&#x27;ve been automating things for ages before Gen Ai came on to the scene.
        • arcanemachiner11 minutes ago
          Yes, it is also a device used to control the movement of work animals, which farmers have been using for ages before QA came on to the scene.
  • fred_tandemai51 minutes ago
    Anyone managed to run pi in a completely sandboxed environment? It can only access the cwd and subdirectories
    • rcarmo24 minutes ago
      I run mine inside <a href="https:&#x2F;&#x2F;github.com&#x2F;rcarmo&#x2F;agentbox" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;rcarmo&#x2F;agentbox</a> (with <a href="https:&#x2F;&#x2F;github.com&#x2F;rcarmo&#x2F;webterm" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;rcarmo&#x2F;webterm</a>)
    • monkey2623 minutes ago
      I do this with an extension. I run all bash tools with bwrap and ACLs for the write and edit tools. Serves my purposes. Opens up access to other required directories, at least for git and rust.<p>I think I published it. Check the pi package page.
    • fjk37 minutes ago
      I’ve been tinkering with Gondolin, a micro-vm agent sandbox.<p>Here’s an example config: <a href="https:&#x2F;&#x2F;github.com&#x2F;earendil-works&#x2F;gondolin&#x2F;blob&#x2F;main&#x2F;host&#x2F;examples&#x2F;pi-gondolin.ts" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;earendil-works&#x2F;gondolin&#x2F;blob&#x2F;main&#x2F;host&#x2F;ex...</a>
  • agentwyz26 minutes ago
    [dead]
  • prakashrj51 minutes ago
    [dead]
  • cyanydeez1 hour ago
    The backing to OpenClaw&#x2F;MoltBot whatever they&#x27;re calling themselves. Why is it insecure, well, Pi tells you &gt;No permission popups.<p>Anyway, even if you give your agent permission, there&#x27;s no secure way to know whether what they&#x27;re asking to is what they&#x27;ll actually do, etc.
    • himata41131 hour ago
      you want to put agents in a sandbox instead such as bwrap anyways.
  • cermicelli1 hour ago
    Just how expensive was that domain?
    • jotaen1 hour ago
      README on Github says “pi.dev domain graciously donated by exe.dev” (though that doesn’t say anything about the original price of course).
      • schpet1 hour ago
        oh that&#x27;s kind. i hope they keep the old domain up too though: <a href="https:&#x2F;&#x2F;shittycodingagent.ai&#x2F;" rel="nofollow">https:&#x2F;&#x2F;shittycodingagent.ai&#x2F;</a>