Lonely Assistant is a small Obsidian plugin that provides an in-vault AI assistant integration focused on Ollama (local and self-hosted Ollama servers). It plans to support other OpenAI-compatible APIs in the future. The plugin makes it simple to use GPT-OSS and other models inside your Obsidian vault with configurable prompts, models, and safety defaults.
- Fast, local-first AI flows using Ollama
- Easy configuration for model, temperature, and token limits
- Minimal, unobtrusive UI inside Obsidian
- Extensible to other OpenAI-compatible endpoints
- Send selected text or the current note to an Ollama model and insert or preview the response
- Per-vault settings for Ollama host, model, temperature, max tokens, and default prompt
- Simple command palette actions and an optional ribbon icon
- Local-first behavior when Ollama is available; configurable fallback for external APIs
- Config validation and sensible defaults
This project is licensed under the MIT License - see the LICENSE file for details.