Methodology

Health is evidence. Fit depends on the job.

ThingsO keeps source facts, AI inference and editorial review separate. The Project Health Score is deterministic and versioned; it is not a universal judgement of software quality.

Source facts

Repository identity, activity, stars, forks, language, license and other factual metadata come from verified source APIs or deterministic processing.

AI inference

Use cases, limitations, fit reasons and relationship candidates are generated from bounded evidence, labelled as inference, and must pass an independent review gate before becoming public-current analysis.

Fit vs health

A healthy project can still be a poor fit for a specific build. Search and comparison therefore treat relevance/fit separately from project health.