# alphaXiv/OpenResearch

Source: [GitHub](https://github.com/alphaXiv/OpenResearch)  
Feed7 permalink: https://feed7.dev/p/openresearch-1472dzs  
Published: Unknown  
Trust: Needs Review (needs_review)

## Why Included

OpenResearch gives coding agents isolated worktrees, reproducible experiment lineage, and local or remote execution for autonomous research loops.

## Source Summary

OpenResearch is a **local-first workspace** that lets Claude Code, Codex, Cursor, or OpenCode run research sessions in **isolated git worktrees**. Its experiment tree links each run to an immutable archive of the recorded commit, alongside logs, diffs, results, and artifacts.

## Practical Implication

Use separate sessions for competing hypotheses, then compare their evidence without losing source lineage. The same committed snapshot can run locally or across SSH, Slurm, Kubernetes, Ray, Hugging Face Jobs, Modal, Tinker, or managed compute.

## Agent-Ready Context

OpenResearch is a **local-first workspace** that lets Claude Code, Codex, Cursor, or OpenCode run research sessions in **isolated git worktrees**. Its experiment tree links each run to an immutable archive of the recorded commit, alongside logs, diffs, results, and artifacts.

Use separate sessions for competing hypotheses, then compare their evidence without losing source lineage. The same committed snapshot can run locally or across SSH, Slurm, Kubernetes, Ray, Hugging Face Jobs, Modal, Tinker, or managed compute.

The dashboard binds to **127.0.0.1**, but a remote service has no application-level authentication, so host access controls matter. Windows support is beta, and official builds send opt-out coarse usage events.

## Connected Context

Feed7 judgment across 760 accumulated Signals:

This makes isolated worktrees and immutable experiment snapshots a concrete local-first harness for comparing research hypotheses across agents and compute backends. It strengthens prior claims that parallelism needs durable evidence, lineage, and isolation, while adding a research-specific experiment tree. It does not establish result quality, and remote use introduces an access-control gap that host security must cover.

- [huangruiteng/loopx](https://feed7.dev/p/loopx-0j0o7ux) — LoopX supplies complementary durable goals, gates, ownership, and quotas, while OpenResearch specializes the persistence layer around immutable research runs, artifacts, and competing hypotheses.
- [Multiplayer agentic engineering — Arjun Singh, Superconductor](https://feed7.dev/p/multiplayer-agentic-engineering-arjun-singh-superconductor-01w4k91) — Both treat isolated sessions and shared visibility as prerequisites for parallel agent work; OpenResearch adds commit-bound experiment lineage and portable compute execution.
- [Don't Build Agents You Can't Answer For — Addy Osmani](https://feed7.dev/p/don-t-build-agents-you-can-t-answer-for-addy-osmani-1y9nwej) — Its archived commits, logs, diffs, results, and artifacts implement much of Osmani’s requirement that high-throughput agent work remain evidence-backed and answerable.
- [From AI-Assisted to AI-Native: Building a Frontier Development Team — Clare Liguori, AWS](https://feed7.dev/p/from-ai-assisted-to-ai-native-building-a-frontier-development-team-clare-0n539qm) — OpenResearch provides concrete infrastructure for the candidate’s parallel-work recommendation, while preserving the scoped sessions and fast comparison needed to keep concurrent hypotheses manageable.

## Context Map

- Layer: agent
- Domains: research, coding
- Topics: harness-engineering, multi-agent, coding-agents

## Uncertainty

- The dashboard binds to **127.0.0.1**, but a remote service has no application-level authentication, so host access controls matter. Windows support is beta, and official builds send opt-out coarse usage events.

## Agent Instruction

Use this item as source-backed context. Do not invent claims beyond the linked source. If this item conflicts with another source, call out the conflict.