timothepoznanski/poznote 6.35.1 on GitHub

Poznote 6.35.1

This patch release polishes the AI Assistant introduced in 6.35.0, with two bug fixes and a dedicated documentation guide.

Stop button now really stops the AI
Clicking stop during an AI chat response only closed the browser connection, while the AI server (e.g. Ollama) kept generating in the background, wasting CPU until the answer was fully produced. The server request is now cancelled within seconds of pressing stop, even while the model is thinking or calling tools.
No more scrollbar arrows in the chat input
On Windows, typing in the AI chat input could make small scrollbar arrows appear next to the text. The input now grows silently up to its maximum height, and only shows a scrollbar for long multi-line messages.

New AI Assistant guide
The AI Assistant now has its own documentation page, docs/AI-ASSISTANT.md, covering providers, choosing a model with tool support, and a full walkthrough of connecting a local Ollama or LM Studio server from the Poznote container: why localhost does not work, how to find the right URL, and how to configure OLLAMA_HOST safely. The README section links to it, following the same pattern as the MCP Server.