Directory Policy
Editorial Methodology
How nTropi qualifies listings, conducts evidence-based assessments, and publishes deterministic software rankings.
What We Rank
nTropi uses one canonical catalog. Real-Coded Apps include published apps whose creators declare a traditional development method. Vibe-Coded Apps include published apps whose creators declare an AI-first method. These labels are self-reported and link back to an official creator statement in our editorial evidence. We do not infer a development method from appearance, code quality, or marketing style. Hybrid and undeclared apps remain visible in the non-ranked Apps index.
Editorial Assessment
App assessments weigh usefulness, usability, reliability, trust, and value equally. Library assessments weigh API design, documentation, reliability, maintenance, and ecosystem fit. Alternative assessments are specific to one target and weigh suitability, feature parity, migration ease, value, and trust.
Rank Calculation
The persisted final score is 70% editorial and 30% Bayesian-adjusted community score. Before public reviews launch, every listing receives the same fixed community prior of 3.5 out of 5. Sponsorship never changes a score or rank.
Ties resolve by editorial score, visible review count, and then listing name. Rankings are rebuilt as a complete context and stored as integer ordinals. Filtering keeps those original ordinals instead of renumbering a subset.
Visible Evidence
Ranked pages publish the ordinal, editorial rationale, and assessment date. Internal rubric criteria, moderation notes, audit records, and score internals are not public.
Sources and Review
Product facts come from official websites, documentation, repositories, repository license files, and creator statements. Summaries and rationales are written by nTropi rather than copied from directories. Each seeded record stores its source URLs and verification timestamp; operational URL checks run outside public requests and temporary failures are sent back for editorial review.
Alternatives
Alternatives are directional: a product can be suitable for one target without being universally better. A public alternatives page requires at least three approved, assessed candidates. Thin or empty comparison pages are withheld.