Cly is entering open beta

[Frostand]

Cly is now public under the Apache-2.0 license. It is a local-first desktop workspace for keeping a computational research question connected to its sources, experiments, local analysis results, claims, and reproducibility review.

The current beta supports deterministic cross-validated binary classification or numeric regression on de-identified numeric CSV/TSV data. It preserves configuration, checksums, metrics, baselines, coefficients, limitations, evidence links, and provenance. The research path works without AI; optional coding sessions can use already authenticated Codex, Claude Code, OpenCode, or Cursor CLI sessions.

Current boundaries are deliberate: no sensitive or regulated data, causal or clinical claims, categorical-predictor handling, notebook import, or hosted sync.

Explore the interactive workspace at cly.one, then follow the beta testing guide or run Cly from source.

Please report the first reproducible place where a real, non-sensitive research workflow breaks the model. Bug reports belong in Issues; workflow feedback and design questions are welcome here.