What it does
An offline-first research engine with declared study protocols, bounded optimization experiments, separate development and holdout seeds, integrity manifests and deterministic replay.
- Declare the question, budget, seeds and acceptance threshold first.
- Evaluate approved search methods on synthetic benchmarks.
- Save reports and independently verify feasibility, accounting and replay.
Start with an inspectable example
- Review a supplied study JSON and the declared limits.
- Run the offline quick start into a new experiment directory.
- Verify the saved run and inspect its report and dashboard.
rei-scientist run examples/sphere_study.json --out runs/my-study
rei-scientist verify runs/my-studyUse the current repository instructions for setup, dependencies and version-specific commands.
Scope and verification
Reference results concern specified synthetic studies. Optional LLM and bibliographic integrations are documented separately; this is not autonomous field-equipment control or independent scientific replication.
Reviewed against the public project documentation on 9 November 2026. The source repository remains the reference for releases, licensing and supported features.