OpenResearch is a local-first workspace designed to transform coding agents—such as Claude Code, Codex, OpenCode, Cursor, and Google Antigravity—into autonomous research agents capable of literature review, hypothesis development, and experimental execution. The framework enables parallel exploration by assigning independent agent sessions to different research directions, utilizing isolated git worktrees to maintain organization. To ensure scientific rigor, the system implements a git-native experiment tree where every run is linked to an immutable archive of its recorded commit, preserving the evidence and context of the results produced. Users can execute workloads locally, via SSH, or through distributed orchestration tools including Slurm, Kubernetes, Ray, Hugging Face Jobs, and Modal. The infrastructure supports various local models through integrations with LM Studio, oMLX, Ollama, or custom endpoints via OpenCode. OpenResearch operates on a local loopback address (127.0.0.1) using a SQLite store to ensure data sovereignty, keeping projects, logs, and artifacts on the user's machine. While it offers managed compute via openresearch.sh, the core environment remains local, allowing users to bridge local development with remote GPUs using the orx up --remote command. The CLI, managed via orx , facilitates complex workflows including project management, experiment execution, and direct paper lookup via arXiv IDs or DOIs.