allenhutchison/obsidian-gemini

What it solves

Gemini Scribe is an Obsidian plugin that transforms your personal knowledge base into an active AI-powered workspace. It solves the problem of AI being disconnected from your personal notes, allowing an AI agent to read, write, and organize your vault while providing context-aware assistance for writing and research.

How it works

The plugin integrates with Google's Gemini models (cloud), Ollama (local), or OpenAI (cloud/compatible servers). It uses a tool-calling agent that can interact directly with the Obsidian vault—searching for files, editing content, and creating new notes. It also implements semantic search via Google's File Search API to find notes by meaning rather than keywords. For automation, it includes lifecycle hooks that trigger AI actions based on vault events (like file creation) and a scheduler for recurring tasks.

Who it’s for

Obsidian users who want to integrate high-powered AI agents into their note-taking workflow to automate vault maintenance, conduct deep research across their notes, or get real-time IDE-style text completions.

Highlights

• Agentic Vault Management: An AI agent capable of searching, reading, creating, and editing notes with a diff-review system for safety.
• Semantic Search: Background indexing of the vault to enable meaning-based retrieval of notes and attachments.
• Multimodal Support: Ability to paste images into chat for analysis and support for SVG rasterization.
• Automation Tools: Lifecycle hooks for event-driven AI runs and a cron-like scheduler for recurring tasks.
• ** uma MCP Server Support**: Experimental integration with the Model Context Protocol to extend agent capabilities with external tools.
• Local & Cloud Flexibility: Support for local models via Ollama or cloud models via Gemini and OpenAI.
• IDE-Style Completions: Real-time, context-aware text suggestions as you type.