3 Commits

Author	SHA1	Message	Date
Jake	3eb24f2c63	docs(phase9): plan corrections from critical review ... Pre-execution review against the actual codebase (Codex unavailable this session) surfaced three mismatches: kon-llm uses LlmEngine with a synchronous generate(prompt, config), AppState exposes llm_engine as a direct Arc not behind RwLock, and llmTags persistence requires a real SQLite migration plus Tauri command extension because saveHistory() is a no-op stub today (which also means manualTags edits weren't persisting). Plan now adds Task 8.5 for migration v14 + storage and Tauri-command extension. Spec data-model section corrected to match. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>	2026-04-24 23:46:06 +01:00
Jake	49a795f533	docs(phase9): design spec for polish debt ... Covers six roadmap items plus Phase 8 carryover backlog: save dialog, bulk export, LLM content tags, progressive-disclosure Settings, visual polish, accessibility sweep. Decisions locked 2026-04-24: on-demand stored LLM tags with grammar-constrained topic + intent closed set; Settings regrouped into one always-expanded "Start here" plus six collapsed details groups; polish and a11y kept to bounded checklists. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>	2026-04-24 22:11:53 +01:00
Jake	2cc0697de9	docs(phase8): design spec for forgiving gamification ... Covers today's completion count badge + 7-day momentum sparkline on the Tasks header. Chose to exclude auto-cascade parents from the count via a new auto_completed column (migration v13), to respect the zero-loss framing. Sparkline is a new settings toggle, default on. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>	2026-04-24 19:53:52 +01:00