<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Nishant Nakum</title>
    <description>The latest articles on DEV Community by Nishant Nakum (@justnishh).</description>
    <link>https://dev.to/justnishh</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3943985%2Fa9d8ad06-2878-4e1c-9093-cd9685c74a33.jpeg</url>
      <title>DEV Community: Nishant Nakum</title>
      <link>https://dev.to/justnishh</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/justnishh"/>
    <language>en</language>
    <item>
      <title>I built a "brain" for AI coding agents — it never forgets and never stops</title>
      <dc:creator>Nishant Nakum</dc:creator>
      <pubDate>Thu, 21 May 2026 11:39:36 +0000</pubDate>
      <link>https://dev.to/justnishh/i-built-a-brain-for-ai-coding-agents-it-never-forgets-and-never-stops-5b54</link>
      <guid>https://dev.to/justnishh/i-built-a-brain-for-ai-coding-agents-it-never-forgets-and-never-stops-5b54</guid>
      <description>&lt;p&gt;I got tired of re-explaining context to my AI every single prompt. It forgets what it did 5 minutes ago. It stops after one step and waits. You end up&lt;br&gt;
  babysitting it more than coding yourself.&lt;/p&gt;


&lt;p&gt;So I built &lt;strong&gt;Long-Horizon&lt;/strong&gt; — a skill that turns any AI coding tool into an autonomous agent with a persistent graph memory.&lt;/p&gt;


&lt;h2&gt;What it does&lt;/h2&gt;


&lt;ol&gt;

  &lt;li&gt;
&lt;strong&gt;You give it one task&lt;/strong&gt; — "Build a REST API with auth"&lt;/li&gt;

  &lt;li&gt;
&lt;strong&gt;It decomposes into subtasks&lt;/strong&gt; and executes them in a loop — no prompting between steps&lt;/li&gt;

  &lt;li&gt;
&lt;strong&gt;It writes knowledge to disk&lt;/strong&gt; as a connected graph — decisions, lessons, patterns, all linked&lt;/li&gt;

  &lt;li&gt;
&lt;strong&gt;It resumes perfectly&lt;/strong&gt; after any interruption — reads its own state and picks up exactly where it stopped&lt;/li&gt;

  &lt;/ol&gt;


&lt;h2&gt;The graph brain&lt;/h2&gt;


&lt;p&gt;Every decision, lesson, and pattern becomes a node. Nodes connect via typed edges (leads_to, caused_by, learned_from). Over time, the AI builds a web&lt;br&gt;
  of project knowledge it can traverse for context.&lt;/p&gt;


&lt;pre&gt;&lt;code&gt;&lt;br&gt;
           ┌─────────┐&lt;br&gt;
      ┌────│  ROOT   │────┐&lt;br&gt;
      │    │(project)│    │&lt;br&gt;
      ▼    └─────────┘    ▼&lt;br&gt;
  ┌────────┐         ┌────────┐&lt;br&gt;
  │decision│─leads─▶ │  task  │&lt;br&gt;
  │"Use JWT"│        │"Build  │&lt;br&gt;
  └───┬────┘         │ API"   │&lt;br&gt;
      │              └───┬────┘&lt;br&gt;
      ▼                  ▼&lt;br&gt;
  ┌────────┐         ┌────────┐&lt;br&gt;
  │ lesson │◀─related─│pattern │&lt;br&gt;
  └────────┘         └────────┘&lt;br&gt;
  &lt;/code&gt;&lt;/pre&gt;


&lt;h2&gt;Live viewer&lt;/h2&gt;


&lt;p&gt;Real-time cyberpunk visualization that shows the brain growing as the AI works:&lt;/p&gt;


&lt;ul&gt;

  &lt;li&gt;Neon glowing nodes with particle trails along edges&lt;/li&gt;

  &lt;li&gt;Unlimited zoom/pan — scroll to zoom, drag to navigate&lt;/li&gt;

  &lt;li&gt;Sound effects when new nodes appear&lt;/li&gt;

  &lt;li&gt;Runs on localhost:3333&lt;/li&gt;

  &lt;/ul&gt;


&lt;h2&gt;Zero dependencies, ~38KB&lt;/h2&gt;


&lt;p&gt;No vector DB. No cloud. No API keys. Pure Node.js, filesystem only. Works offline.&lt;/p&gt;


&lt;h2&gt;Works with everything&lt;/h2&gt;


&lt;p&gt;One command installs for your tool:&lt;/p&gt;


&lt;pre&gt;&lt;code&gt;npx long-horizon init&lt;br&gt;
  npx long-horizon adapt cursor    # or: windsurf, claude, aider, codex, all&lt;br&gt;
  &lt;/code&gt;&lt;/pre&gt;


&lt;p&gt;Also ships an MCP server with 11 tools for direct AI integration.&lt;/p&gt;


&lt;h2&gt;It's always-on&lt;/h2&gt;


&lt;p&gt;Once installed, you don't need to say "use Long-Horizon" or prefix your prompts. The AI reads the skill file and operates autonomously by default. Just&lt;br&gt;
  talk normally:&lt;/p&gt;


&lt;pre&gt;&lt;code&gt;Build a user dashboard with charts and dark mode.&lt;/code&gt;&lt;/pre&gt;


&lt;p&gt;The AI loops until done.&lt;/p&gt;


&lt;h2&gt;Links&lt;/h2&gt;


&lt;ul&gt;

  &lt;li&gt;
&lt;strong&gt;GitHub:&lt;/strong&gt; &lt;a href="https://github.com/justnishh/long-horizon" rel="noopener noreferrer"&gt;github.com/justnishh/long-horizon&lt;/a&gt;
&lt;/li&gt;

  &lt;li&gt;
&lt;strong&gt;Install:&lt;/strong&gt; &lt;code&gt;npx long-horizon init&lt;/code&gt;
&lt;/li&gt;

  &lt;li&gt;
&lt;strong&gt;License:&lt;/strong&gt; MIT&lt;/li&gt;

  &lt;/ul&gt;


&lt;p&gt;Would love feedback. What would make this more useful for your workflow?&lt;/p&gt;

</description>
      <category>ai</category>
      <category>opensource</category>
      <category>productivity</category>
      <category>programming</category>
    </item>
  </channel>
</rss>
