Changelog

Added deterministic normalized lexical prefix completion through Lexicon.complete() and the complete CLI command

Added automatic color styling for interactive human dictionary output with explicit opt-out controls

Added deterministic local lexical prefix completion through Lexicon.complete() and lexhint complete

Changed WordEvidence to expose lexical case attestation and uppercase-only evidence

Fixed semantic context anchoring so character-span numeric and symbolic targets do not consume adjacent cues

Added GitHub release dataset discovery and installation for versioned Lexhint SQLite artifacts

Added schema-aware dataset variants and regional locale preferences for dictionary lookup and rendering

Added CLI commands for lookup, segmentation, context evidence, artifact builds, dictionary inspection, and status

Added streamed Wiktextract and Kaikki dictionary builds that produce self-describing SQLite artifacts

Added stable semantic-domain evidence that explains context cues while excluding the target span

Added a local read-only Lexicon for dictionary-derived membership and compact identifier segmentation

Added zero runtime dependencies and support for Python 3.10 through 3.14

Changed FrequencyWords integration to optional counts and ranks that enrich dictionary-derived lexemes with provenance

Changed rich dictionary entries to preserve senses, forms, pronunciations, examples, relations, and etymology

Removed standalone word-list vendoring in favor of dictionary-derived lexical membership

Removed lazy runtime dictionary fetching and per-word artifact population

Documented the local SQLite artifact architecture and Sphinx documentation setup

Improved CI and publication checks for tests, static analysis, documentation, distributions, and wheel smoke tests