pfrankov/obsidian-local-gpt
Local Ollama and OpenAI-like GPT's assistance for maximum privacy and offline access
What it solves
It provides a way to integrate local AI assistants into the Obsidian note-taking app, allowing users to perform AI-powered text manipulation and analysis without leaving their workspace. It eliminates the need to copy-pasting text between Obsidian and external AI interfaces.
How it works
The plugin connects to AI providers (via the AI Providers plugin) to run models. It offers a variety of predefined and community-created "actions"—such as summarizing text, fixing grammar, or continuing writing—which can be triggered via a context menu or an Action Palette. It supports multimodal capabilities (images) and RAG (Retrieval-Augmented Generation) to pull context from links, backlinks, and PDF files within the Obsidian vault.
Who it’s for
Obsidian users who want to leverage local LLMs for writing assistance, note organization, and knowledge management within their personal knowledge base.
Highlights
- RAG Support: Uses embedding models to pull context from links, backlinks, and PDFs.
- Multimodal: Supports AI actions that work with images.
- ** uma la l'Action Palette**: A searchable interface for running one-time AI requests.
- Community Actions: Ability to browse and install shared AI prompts/actions directly within the plugin.
- Reasoning Modes: Allows users to select specific reasoning modes and creativity (temperature) settings for different models.
- Multilingual: Supports a wide range of global languages.
Related
- Project
- Project
- Project
- Project