a claude code/codex plugin to run autoresearch on your repository

Dev.to / 4/18/2026

💬 OpinionDeveloper Stack & InfrastructureSignals & Early TrendsTools & Practical Usage

Key Points

  • The article introduces an open-source Claude Code/Codex plugin called Evo that performs “autoresearch” style optimization on a user’s repository using experiments and benchmarks.
  • It takes a codebase, identifies a benchmark, runs a baseline, then launches multiple parallel agents to try to improve performance, keeping only changes that beat the baseline.
  • The plugin structures the autoresearch approach with tree search via greedy hill-climbing, parallel execution using Git worktrees, shared failure traces to reduce repeated mistakes, and regression gates to prevent harmful changes.
  • Each experiment is implemented as a Git worktree branch; improvements that pass scoring and regression gates are committed, while regressions are discarded and cleaned up, with results visible in a local dashboard.
  • Setup is straightforward via the Claude Code plugin marketplace commands and the project is licensed under Apache 2.0 without requiring additional signups or API keys beyond Claude Code itself.

i built

https://github.com/evo-hq/evo

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 & Codex plugin that optimizes code through experiments

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.

inspired by karpathy's autoresearch, but with structure on top:

  • tree search over greedy hill-climb — multiple forks from any committed node
  • N parallel agents in git worktrees
  • shared failure traces so agents don't repeat each other's mistakes
  • regression gates

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

Apache 2.0, no signup, no API keys beyond what Claude Code already has:

/plugin marketplace add evo-hq/evo
/plugin install evo@evo-hq-evo