Ship Assistent 0.10.8 cheap-bug sweep for RU intent, VRAM park, and identity.

Silent Generate, pack selection, and loading UI no longer trip on Cyrillic or fake GPU loads; persona title is the assistant, not the user.

Co-authored-by: Cursor <cursoragent@cursor.com>

README.md

@@ -2,7 +2,7 @@
 
 SwarmUI extension for **collaborative Krea 2** prompting via **Ollama**: chat + board (Generate | Refs tabs), LoRA chips, **persona presets** (`Config/personas/`), **About the user** prefs + craft vector memory, model cards with Civitai fetch, img2img/inpaint, slash commands, auto Generate.
 
-**Version 0.10.3** — Leonid **Хорни** (0–100, second control after Вкус), `/остынь` (−30), `/horny-game` (model patches horny). Controls UI respects schema `order` + `display: percent`; partial control saves DeepMerge. Builds on 0.10.2 context/vision lean pass.
+**Version 0.10.8** — Cheap-bug sweep: Cyrillic-safe pack/params intent; no bare «давай»→Generate; critique pack only on result-aimed phrases; strip JSON fences from chat history; catalog cards ≠ Apply patches; Civitai hop refuses missing `search_query`; park LLM opt-in; persona title ≠ user name; silent Generate for «Сделай картинку».
 
 ## Layout
 
@@ -74,8 +74,8 @@ Two layers in `memory/assistent.sqlite` (`persona` column; empty = shared):
 - `settings.json` and persona overlays stay files (layered merge + git). `.assistent.json` cards stay next to weights.
 ## VRAM handover
 
-- Before every Generate the chat model is unloaded (`keep_alive: 0`) so Krea 2 gets the whole GPU
-- Back in the Chat tab it is warmed again with a 1-token request (`keep_alive 15m`, `num_ctx` from `assistant.json`)
+- Before Generate the chat model is unloaded only if **Park LLM** is enabled (`keep_alive: 0`) — default off so VL chat stays warm; enable when Generate OOMs
+- When parked, after Generate the model is warmed again (`keep_alive 15m`) before the UI goes idle
 - Embed / memory models are never parked — reloading them would stall every retrieve
 
 ## UX