alphaXiv Opens OpenResearch, Turning Coding Agents Into Autonomous Scientists
alphaXiv released a local-first, MIT-licensed CLI that spins up parallel Claude Code, Codex, or OpenCode agents against isolated git worktrees.
PRO- alphaXiv open-sourced OpenResearch, a local-first Rust workspace for parallel research agents, under MIT license.
- Wraps Claude Code, Codex, or OpenCode with isolated git worktrees and a git-native experiment tree.
- Autoresearch mode runs the full propose, code, experiment, inspect loop autonomously.
- Runs locally, over SSH, or on Slurm, Kubernetes, Ray, Hugging Face Jobs, Modal, and Tinker.
- New installable skills system includes a figures skill for publication-quality visualizations.
- Install via
curl -LsSf https://openresearch.sh/install.sh | shthenorx up.
alphaXiv has open-sourced OpenResearch, a Rust-based workspace that turns coding agents into autonomous research agents. It runs locally, keeps code and data on your machine, and fans out multiple agent sessions in parallel across independent git worktrees so different hypotheses stay isolated.
The target workflow: point Claude Code, Codex, or OpenCode at a research problem and let it review literature, develop hypotheses, run experiments, and produce research artifacts. You pick the harness and model per session; OpenResearch handles orchestration, evidence tracking, and compute routing.
What Ships in the Box
OpenResearch comes as both a desktop app and a CLI called orx. Installation on macOS or Linux is a one-liner:
curl -LsSf https://openresearch.sh/install.sh | sh
orx uporx up opens a local dashboard at http://127.0.0.1:4791, backed by SQLite on loopback, so nothing leaves your machine when you create a project or launch a run. The core primitives:
- Parallel exploration where each research direction gets an independent agent session and isolated git worktree.
This story is for Pro members
You've reached the end of the free preview. Upgrade to AlphaSignal Pro to read the full article - and everything else behind the paywall.