The latest articles on DEV Community by 승완 양 (@__4cbbf6020a1e29ec6258). https://dev.to/__4cbbf6020a1e29ec6258 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%2F3875037%2Fa1686d43-5b34-4472-8b40-1f7e17db9383.png https://dev.to/__4cbbf6020a1e29ec6258 en 승완 양 Sun, 12 Apr 2026 14:55:07 +0000 https://dev.to/__4cbbf6020a1e29ec6258/i-built-simpleralph-a-file-driven-autonomous-coding-loop-for-any-repo-5hb https://dev.to/__4cbbf6020a1e29ec6258/i-built-simpleralph-a-file-driven-autonomous-coding-loop-for-any-repo-5hb <h1> I built SimpleRalph — a file-driven autonomous coding loop for any repo </h1> <p>I just open-sourced <strong>SimpleRalph</strong>:</p> <p><a href="https://github.com/Seungwan98/SimpleRalph" rel="noopener noreferrer">https://github.com/Seungwan98/SimpleRalph</a></p> <p>SimpleRalph is a lightweight CLI for running a <strong>file-driven autonomous coding loop</strong> inside any repository.</p> <p>The core idea is simple:</p> <ul> <li>give it one topic</li> <li>create a local session under .simpleralph/</li> <li>keep the loop state explicit with <strong>PRD / Tasks / Status / Log</strong> </li> <li>run compile/test gates between iterations</li> <li>export artifacts when needed</li> </ul> <h2> Why I made it </h2> <p>A lot of agent workflows feel either:</p> <ol> <li>too chat-memory-driven to resume cleanly</li> <li>too opaque to inspect when something goes wrong</li> </ol> <p>I wanted something more explicit and reviewable.</p> <p>SimpleRalph keeps the working state on disk instead of burying everything in one long conversation.</p> <h2> Current commands </h2> <ul> <li><code>simpleralph init</code></li> <li><code>simpleralph run</code></li> <li><code>simpleralph status</code></li> <li><code>simpleralph export</code></li> </ul> <h2> A small detail I cared about </h2> <p>If the target repository contains an <code>AGENTS.md</code>, the default session config makes that file available to the agent run automatically.</p> <p>The package is <strong>agent-agnostic</strong> at the config level, even though the default generated command template is OpenCode-friendly.</p> <h2> Status </h2> <p>It is still alpha, but the current GitHub version has already been tested from a fresh clone with:</p> <ul> <li>install</li> <li>init</li> <li>status</li> <li>export</li> <li>one safe loop run</li> </ul> <p>If you try it, I would love feedback on:</p> <ul> <li>whether the file-driven model feels useful</li> <li>where the UX feels too heavy</li> <li>which agent CLIs should be better supported first</li> </ul> productivity