green-dalii/obsidian-llm-wiki

What it solves

Many users have vast collections of notes in Obsidian that become difficult to navigate as they grow. Traditional search or folder structures often fail to surface non-obvious connections between ideas. This plugin automates the creation of a structured, interconnected wiki from these notes, allowing users to chat with their knowledge base and discover relationships without manually tagging or organizing every file.

How it works

The plugin uses an LLM to extract entities (people, organizations) and concepts (theories, terms) from your notes, creating standalone wiki pages for each. Instead of using a vector database or embeddings, it employs a graph-based retrieval system using Personalized PageRank (PPR) and Monte Carlo simulations over the [[wiki-link]] graph. This allows the AI to retrieve relevant context based on the structural connectivity of your notes rather than just semantic similarity.

Who it’s for

• Obsidian users who want a "second brain" that is queryable via AI.
• Privacy-conscious users who prefer local-first setups (via Ollama or LM Studio) to keep data offline.
• Non-technical users who want a one-click installation without managing Python environments or vector databases.
• Researchers who need to ingest PDFs, images, and Office documents into their knowledge graph.

Highlights

• Zero-Embedding Architecture: Uses graph-based retrieval (PPR) instead of a vector DB, removing the need for embedding models.
• Broad Provider Support: Compatible with 16+ LLM providers, including OpenAI, Anthropic, Gemini, and local options like Ollama.
• Multi-Format Ingest: Supports Markdown, PDF, images, and Office files (DOCX, PPTX, XLSX) via MinerU or native cloud provider PDF support.
• Integrated Workflow: Runs entirely within Obsidian, utilizing the native Graph View for visualization and a right-docked side panel for chatting.
• Bilingual/Multilingual: Supports 11 native languages for both the user interface and the generated wiki output.