YazSes v2.7.0-dev.1 — Wave K opens (developer preview)¶

Pre-release · not published (publish workflows are pinned to v0.x/v1.x; v2 dev tags do not ship). v1.4.1 remains the stable release. Every feature is OFF by default.

Opens Wave K — a fresh 2026 SoA round (design/vision/v2-research/13-wave-k.md) of 10 features distinct from the prior 82, prioritized for accessibility and things a cloud tool structurally can't do (they need local keyboard/window/selection/clipboard state). This first drop ships the two cleanest fully-pure cores:

New features (all off by default)¶

• Chorded Shortcut Synthesis (ADR-v2-085, [chords]) — say any keyboard shortcut and it's pressed: "press control shift P", "escape twice", "hit F5" — no macro registration needed. parse_chord (modifiers + named/function keys + repeat) reuses the existing injector, no new backend. The missing "keyboard-shortcut layer" and the biggest motor-accessibility win of the wave.
• Inline Compute (ADR-v2-086, [compute]) — "what's 15% of 240" → types 36. On-device arithmetic and percentages evaluated through a safe AST walker (never Python eval), so a misrecognition can't execute code. Distinct from Voice Unit Conversion (converts), Spoken Math→LaTeX (formats), and the Temporal Normalizer (normalizes) — this evaluates. Privacy-clean for sensitive figures. Algebra (sympy) and date math are deferred.

yazses features now lists 97 capabilities.

Wave K backlog¶

Focus-Class Auto-Profile Switching, Voice Undo/Redo Timeline, Voice Case/Identifier Transform, Auto-Pairing & Wrap-Selection, Spoken Table→CSV Data Entry, Session Bookmarks & Resume, Word-Count/ Goal Tracker, Local Voice Timer/Break Reminder.

Quality¶

1306 tests green across Linux × macOS × Windows on Python 3.11 and 3.12. Both Wave K cores are unit-tested to 100% with no new dependency; the default install and the v1 dictation path are unchanged.