Changelog

Added SSMD-aware reporting, context lookup, TXT projection, and safe speech audit and write-back workflows

Fixed platform-independent analysis freshness by normalizing SSMD newlines and preserving workspace byte hashes

Fixed bundle workspace safety and newline-stable source checks during speech materialization

Added native SSMD scene-break handling for Markdown, HTML, and EPUB conversions

Added adaptive CLI help, portable metadata overrides, book language selection, and bundle metadata inspection

Changed the default for new conversions to sequence_fallback_mode=preserve while retaining explicit spell support

Added in-memory conversion of text, Markdown, and HTML, with Markdown rendered as SSMD-safe spoken text

Added spell or preserve sequence fallback modes to document and book conversion, saved in SSMD and book metadata

Documented the boundary between ssmdconvert’s conversion role and ttsready’s downstream speech review

Added EPUB chapter inspection, source-number selection, and standalone SSMD chapter conversion

Added editable directory and portable ZIP bundles with manifest-defined chapter order

Added case-insensitive .ssmd.md detection as an SSMD input suffix

Changed EPUB chapter extraction to use epub2text while preserving combined conversion results

Added dedicated book inspect, convert, and validate CLI subcommands with JSON inspection and validation output

Made book bundle formats explicit and writes atomic, deterministic, and non-overwriting unless requested

Recorded source hashes at book conversion and separated original navigation parents from canonical chapter parents

Moved speech preparation dependencies into the optional speech extra