‹ BackHN Continuity

Thread

Pi 1.0

1684 points · 602 comments · sergiotapia

  1. utilize1808 · · focus · HN ↗
    I don't understand why "Cache warming for anthropic models" was not put into a standalone package, but has to be bundled with the "minimal" coding agent.
    1. threecheese · · focus · HN ↗
      It's table stakes for using Anthropic APIs cost-consciously. They aren't building a minimal harness for nerd points; they are building it to be usable, and dogfooding it with Real Money to learn what that means realistically.

      This might've made a better argument ca. 2005 wrt a 20-something Austrian's minimalist wsgi framework. Heck, I'll bet you could even find said argument somewhere in the Archive :)

      1. mixermachine · · focus · HN ↗
        Copied from myself above:

        That goes against the philosophy of pi. Nearly Everything is a Plugin. Need to set the reasoning effort? Install pi-reasoning. Need subagents? Install pi-subagents. Need permission gating? Install one of the permission extensions (otherwise the agent can do everything).

        A feature like cache warming on Anthrophic feels very strange in this environment. NOT saying that it is not useful. It just should be a plugin.

        1. kelnos · · focus · HN ↗
          Changing the reasoning effort is built in and doesn't require a plugin.
          1. mixermachine · · focus · HN ↗
            Yer, that is my bad for not updating myself, sorry. I did miss the introduction of Alt + Tab or /thinking command for switching this.

            Good thing I can remove another package.

        2. endinator · · focus · HN ↗
          That's your philosophy of pi and it seems you are taking it from deepseek. It's stated that pi wants to keep the agent loop minimal, not the code around the harness. Also those extensions in pi can sometimes be very bad, as it depends on the maintainer. And when it comes to cache handling of the best frontier coding model, it's much better that pi itself maintains it rather than some rando online.
          1. mixermachine · · focus · HN ↗
            No I did not ask AI for this. This is the understanding I have from working with PI for more than a year.

            They also say so on their website ( <a href="https:&#x2F;&#x2F;pi.dev&#x2F;" rel="nofollow">https:&#x2F;&#x2F;pi.dev&#x2F; ): &quot;Pi ships with powerful defaults but skips features like sub-agents and plan mode. Ask Pi to build what you want, or install a package that does it your way.&quot;

    2. ruined · · focus · HN ↗
      because &quot;i came back from lunch and my fresh 5h usage limit evaporated&quot; is not a feature

      i guess you could blame the api, if you wanted.

      1. utilize1808 · · focus · HN ↗
        What 5h usage limit?

        On a more serious note: pi-agent shouldn&#x27;t know about such arbitrary limits imposed by a company that gets paranoid when people use thrid-party agents to consume their Claude subscriptions.

        1. shepherdjerred · · focus · HN ↗
          It’s not reasonable to ignore one of the biggest players in the space no matter how much you disagree with them
          1. what · · focus · HN ↗
            It is actually reasonable to ignore them. It should be plugin&#x2F;extension you can install if you want.
          2. mixermachine · · focus · HN ↗
            That goes against the philosophy of pi. Nearly Everything is a Plugin. Need to set the reasoning effort? Install pi-reasoning. Need subagents? Install pi-subagents. Need permission gating? Install one of the permission extensions (otherwise the agent can do everything).

            A feature like cache warming on Anthrophic feels very strange in this environment. NOT saying that it is not useful. It just should be a plugin.

          3. utilize1808 · · focus · HN ↗
            I not saying Pi should ignore them. I am saying that for a coding agent that has been advocating leanness and minimalism and &quot;everything is a plugin&quot;, it&#x27;s outlandishly bazaar for them to include this specific functionality into the agent proper.
        2. jonwinstanley · · focus · HN ↗
          ChatGPT Work mode is the one with a 5 hour usage limit I think
      2. Hamuko · · focus · HN ↗
        Anthropic&#x27;s API has a five-hour usage limit?
    3. tcdent · · focus · HN ↗
      Yeah, it&#x27;s such a subjective feature. I don&#x27;t really worry about the five minute timeout, but sometimes if I&#x27;m conscious about the fact that I have an 800k token context window that I haven&#x27;t interacted with for a day, I will switch to Opus for compaction and then back to Fable for interaction.

      In general, though, some kind of API ping on a timeout does not support my personal workflow, which involves dozens of active or stagnant agent sessions that stay open for weeks.

Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.