The latest articles on DEV Community by Alok Bishoyi (@alokbishoyi97). https://dev.to/alokbishoyi97 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%2F3885376%2Fbdab2271-15b1-4ad7-a7c6-8138cb07e38f.jpg https://dev.to/alokbishoyi97 en Alok Bishoyi Sat, 18 Apr 2026 02:39:50 +0000 https://dev.to/alokbishoyi97/a-claude-codecodex-plugin-to-run-autoresearch-on-your-repository-45oa https://dev.to/alokbishoyi97/a-claude-codecodex-plugin-to-run-autoresearch-on-your-repository-45oa <p>i built</p> <p><a href="https://github.com/evo-hq/evo" rel="noopener noreferrer">https://github.com/evo-hq/evo</a></p> <p>for those of you who are autoresearch pilled , or have been meaning to get into autoresearch but dont know how. Its an opensource Claude Code &amp; Codex plugin that optimizes code through experiments</p> <p>you hand it a codebase. it finds a benchmark, runs the baseline, then fires off parallel agents to try to beat it. kept if better, discarded if worse.</p> <p>inspired by karpathy's autoresearch, but with structure on top:</p> <ul> <li>tree search over greedy hill-climb — multiple forks from any committed node</li> <li>N parallel agents in git worktrees</li> <li>shared failure traces so agents don't repeat each other's mistakes</li> <li>regression gates</li> </ul> <p>under the hood: each experiment is a git worktree branching from its parent. commits on score improvement + gate pass. discards + worktree cleanup on regression. everything observable in a local dashboard</p> <p>Apache 2.0, no signup, no API keys beyond what Claude Code already has:</p> <p><code>/plugin marketplace add evo-hq/evo</code><br> <code>/plugin install evo@evo-hq-evo</code></p> claudecode ai algorithms