11 comments

  • nullbio50 minutes ago
    &gt; Agents propose and publish capabilities to a shared contribution site, letting others discover, adopt, and evolve them further. A collaborative, living ecosystem of personal AIs.<p>While I like this idea in terms of crowd-sourced intelligence, how do you prevent this being abused as an attack vector for prompt injection?
    • adriancooney44 minutes ago
      100%. This is why I&#x27;m so reluctant to give any access to my OpenClaw. The skills hub is poisoned.
    • ddaniel1031 minutes ago
      Great point. I wrote it as important note and ill take it into account.
  • aaaalone2 hours ago
    I will not download or use something which constantly reminds me of this weird dude suckerberg who did a lot of damage to society with facebook
    • 62746724 minutes ago
      This Zuckerman[0] would like a word<p>[0] <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Mortimer_Zuckerman" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Mortimer_Zuckerman</a>
    • ddaniel1027 minutes ago
      Haha it&#x27;s your personal agent, let him handle the stuff you don&#x27;t like. But soon, right now its not fully ready
    • philipallstar1 hour ago
      That&#x27;s really good to know
    • gowld1 hour ago
      Zuckerberg.<p>At first I thought it was a naming coincidence, but looking at the zuckerman avatar and the author avatar, I&#x27;m unsure if it was intentional:<p><a href="https:&#x2F;&#x2F;github.com&#x2F;zuckermanai" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;zuckermanai</a><p><a href="https:&#x2F;&#x2F;github.com&#x2F;dvir-daniel" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;dvir-daniel</a><p><a href="https:&#x2F;&#x2F;avatars.githubusercontent.com&#x2F;u&#x2F;258404280?s=200&amp;v=4" rel="nofollow">https:&#x2F;&#x2F;avatars.githubusercontent.com&#x2F;u&#x2F;258404280?s=200&amp;v=4</a><p>The transparency glitch in GitHub makes the avatar look either robot or human depending on whether the background is white or black. I don&#x27;t know if that&#x27;s intentional, but it&#x27;s amazing.
    • zeroonetwothree1 hour ago
      I was hoping it was a Philip Roth reference but I was disappointed when I opened the page.
  • 4b11b43 hours ago
    DIY agent harnesses are the new &quot;note taking&quot;&#x2F;&quot;knowledge management&quot;&#x2F;&quot;productivity tool&quot;
    • ddaniel103 hours ago
      DIYWA - do it yourself with agent ;) hopefully zuckerman as the start point
  • asim1 hour ago
    I started working on something similar but for family stuff. I stopped before hitting self editing because, well I was a little bit afraid of becoming over reliant on a tool like this or becoming more obsessed with building it than actually solving a real problem in my life. AI is tricky. Sometimes we think we need something when in fact life might be better off simpler.<p>The code for anyone interested. Wrote it with exe.dev&#x27;s coding agent which is a wrapper on Claude Opus 4.5<p><a href="https:&#x2F;&#x2F;github.com&#x2F;asim&#x2F;aslam" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;asim&#x2F;aslam</a>
  • joonate51 minutes ago
    |The agent can rewrite its own configuration and code.<p>I am very illiterate when it comes to Llms&#x2F;AI but Why does nobody write this in Lisp???<p>Isn&#x27;t it supposed to be the language primarily created for AI???
    • lm2846940 minutes ago
      &gt; Isn&#x27;t it supposed to be the language primarily created for AI???<p>In 1990 maybe
    • tines43 minutes ago
      Nah, it’s pretty unrelated to the current wave of AI.
  • lmf4lol18 minutes ago
    I am surprised that no one did this in a LISP yet.
  • ddaniel105 hours ago
    Hi HN,<p>I&#x27;m building Zuckerman: a personal AI agent that starts ultra-minimal and can improve itself in real time by editing its own files (code + configuration). Agents can also share useful discoveries and improvements with each other.<p>Repo: <a href="https:&#x2F;&#x2F;github.com&#x2F;zuckermanai&#x2F;zuckerman" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;zuckermanai&#x2F;zuckerman</a><p>The motivation is to build something dead-simple and approachable, in contrast to projects like OpenClaw, which is extremely powerful but has grown complex: heavier setup, a large codebase, skill ecosystems, and ongoing security discussions.<p>Zuckerman flips that:<p>1. Starts with almost nothing (core essentials only).<p>2. Behavior&#x2F;tools&#x2F;prompts live in plain text files.<p>3. The agent can rewrite its own configuration and code.<p>4. Changes hot-reload instantly (save -&gt; reload).<p>5. Agents can share improvements with others.<p>6. Multi-channel support (Discord&#x2F;Slack&#x2F;Telegram&#x2F;web&#x2F;voice, etc).<p>Security note: self-edit access is obviously high-risk by design, but basic controls are built in (policy sandboxing, auth, secret management).<p>Tech stack: TypeScript, Electron desktop app + WebSocket gateway, pnpm + Vite&#x2F;Turbo.<p>Quickstart is literally:<p><pre><code> pnpm install &amp;&amp; pnpm run dev </code></pre> It&#x27;s very early&#x2F;WIP, but the self-editing loop already works in basic scenarios and is surprisingly addictive to play with.<p>Would love feedback from folks who have built agent systems or thought about safe self-modification.
    • iisweetheartii5 hours ago
      Love the minimalist approach! The self-editing concept is fascinating—I&#x27;ve seen similar experiments where the biggest early failure points are usually:<p>1. Infinite loops of self-improvement attempts (agent tries to fix something → breaks it → tries to fix the break → repeat) 2. Context drift where the agent&#x27;s self-modifications gradually shift away from original goals 3. File corruption from concurrent edits or malformed writes<p>Re: sharing self-improvements across agents—this is actually a problem space I&#x27;m actively working on. Built AgentGram (agentgram.co) specifically to tackle agent-to-agent discovery and knowledge sharing without noise&#x2F;spam. The key insight: agents need identity, reputation, and filtered feeds to make collaborative learning work.<p>Happy to chat more about patterns we&#x27;ve found useful. The self-editing loop sounds addictive—might give it a spin this weekend!
      • junon1 hour ago
        AI generated response on a post about AI. Getting tired of this timeline.
        • ohyoutravel1 hour ago
          Not only that, but the OP created that account solely to hype their own product lol. There’s another bot downthread doing the same thing. Minimally it feels like dang should not let new accounts post for 30 days or something without permission.
          • yborg34 minutes ago
            That might reduce botting for about 30 days, people will just tee up an endless supply of parked ids that will then spin up to post after the lockout expires.
        • anarticle3 minutes ago
          Why not ban both accounts? Seems like a fine way to keep SNR high to me.
        • nullbio48 minutes ago
          Yep. It&#x27;s very obvious, and lazy.
    • ekinertac4 hours ago
      there are hardcoded elements in the repo like:<p>&#x2F;Users&#x2F;dvirdaniel&#x2F;Desktop&#x2F;zuckerman&#x2F;.cursor&#x2F;debug.log
  • falloutx1 hour ago
    Terrible name, kind of a mid idea when you think about it (Self improving AI is literally what everyone&#x27;s first thought is when building an AI), but still I like it.
    • ddaniel1025 minutes ago
      Thanks for the feedback. Are you going to forget this name though?
  • amelius3 hours ago
    Sounds cool, but it also sounds like you need to spend big $$ on API calls to make this work.
    • ddaniel103 hours ago
      I&#x27;m building this in the hope that AI will be cheap one day. For now, I&#x27;ll add many optimizations
      • Zetaphor1 hour ago
        Have you tested this with a local model? I&#x27;m going to try this with GLM 4.7
        • mcny55 minutes ago
          What would be the best model to try something like this on a 5800XT with 8 GB RAM?
      • amelius3 hours ago
        Yes, it certainly makes sense if you have the budget for it.<p>Could you share what it costs to run this? That could convince people to try it out.
        • ddaniel102 hours ago
          I mean, you can just say Hi to it, and it will cost nothing. It only adds code and features if you ask it to
      • croes2 hours ago
        AI is cheap right now. At some point the AI companies must turn to generate profit
  • pillbitsHQ1 hour ago
    [dead]
  • pillbitsHQ4 hours ago
    [dead]