alphaXiv/OpenResearch

Turn your coding agents into research agents

OpenResearch – a local‑first workspace for AI research agents

What it is

  • A desktop (macOS/Windows/Linux) and CLI application that lets you run research agents – LLM‑powered assistants that can read scientific papers, generate hypotheses, execute code, and log results.
  • The tool is part of the alphaXiv ecosystem and is built to keep all data on your own machine (local SQL DB, file storage).

Key capabilities

Feature What it does
Agent integration Works out‑of‑the‑box with popular coding assistants such as Claude, Codex, OpenCode, Cursor, and Google Antigravity. You can also point it at a local model via LM Studio, oMLX, Ollama or any custom endpoint.
Literature grounding Connects to pre‑print servers (alphaXiv, bioRxiv) and PubMed so agents can fetch the latest papers, extract relevant information and base hypotheses on real research.
Compute flexibility Experiments can run locally or be dispatched to remote resources: SSH, Slurm, Kubernetes, Modal, Tinker, Ray, Hugging Face Jobs, or the managed OpenResearch cloud.
Experiment memory Every run – code, logs, artifacts – is stored in a local SQLite database. The desktop UI lets you browse the full history, so agents can “remember” past outcomes and plan next steps.
Privacy‑first No code, prompts or file paths are sent to the vendor. Only anonymised usage counts (model name, token/message volume) are collected, and you can disable telemetry completely.
Cross‑platform Pre‑built installers for macOS (DMG), Windows (Setup.exe, beta), and Linux (AppImage). A lightweight CLI (orx) can also be installed via a one‑liner script.

How to get started

  1. Download the desktop app for your OS from the Releases page (or install the orx CLI with the provided script).
  2. Create an account at openresearch.sh to receive updates and access managed compute resources.
  3. Launch an agent – pick a coding model (Claude, Codex, local Ollama, etc.) and let it browse literature, propose experiments, and run them on your chosen compute backend.
  4. Inspect results in the UI or via the local dashboard (orx up → http://127.0.0.1:4791).

Who it’s for

  • Researchers who want AI assistance for literature review, hypothesis generation, and reproducible experiment tracking.
  • Developers building AI‑augmented scientific workflows who need a plug‑and‑play interface to multiple LLM providers and compute clusters.
  • Anyone who values keeping their research data under personal control while still leveraging powerful language models.

Documentation & support

  • Full docs at openresearch.sh/docs
  • Platform‑specific install guides (Windows, Linux) linked from the README.
  • Community feedback is collected anonymously; you can turn it off in Settings or via orx telemetry off.

OpenResearch is a genuine, actively maintained open‑source project that sits at the intersection of AI agents, large‑language‑model tooling, and reproducible scientific computing.

Related

  • Project
    openteams-lab/openteams

    A local-first AI desktop app that provides a shared workspace for multiple coding agents to collaborate, plan, and execute tasks without interfering with each other.

  • Project
    uditgoenka/autoresearch

    An autonomous improvement engine for AI coding agents that uses a loop of modification, mechanical verification, and git-based rollbacks to optimize code and systems.

  • Project
    mrsibe/KnowNote

    A local-first research workspace that turns documents into a knowledgeable base with traceable AI answers and local RAG processing.

  • Project
    u14app/neo-chat

    A self-hostable, local-first AI workspace that integrates multi-model chat, agentic workflows, and deep research tools with a focus on user data privacy.

  • Project
    OpenRaiser/NanoResearch

    An end-to-end autonomous AI research engine that automates the process from ideation to LaTeX paper production by executing real GPU experiments to ensure data authenticity.