Summary

xoai/sage-wiki addresses the challenge of organizing disparate personal knowledge into a cohesive, searchable wiki. It targets engineers, researchers, and power users seeking to leverage LLMs for knowledge synthesis. The primary benefit is transforming raw documents into an interconnected, intelligently structured knowledge base that compounds over time and integrates with existing workflows.

How It Works

The project implements an LLM-compiled personal knowledge base using the Sage Framework. Users deposit documents (various formats supported) into a designated folder. The system automatically detects formats, processes content via LLM calls for summarization and concept extraction, and generates interconnected wiki articles. This approach enables compounding knowledge, where each new source enriches the existing wiki, making it progressively smarter.

Quick Start & Requirements

• Installation: CLI-only via go install github.com/xoai/sage-wiki/cmd/sage-wiki@latest. Web UI requires cloning the repo, npm install, npm run build, and go build -tags webui.
• Prerequisites: Go, Node.js (for Web UI), LLM API keys (OpenAI, Gemini, Anthropic, Ollama, etc.), and potentially a vision-capable LLM for image analysis.
• Links: sage-wiki init for setup, sage-wiki serve --ui for Web UI, MCP Knowledge Capture Guide (via README).

Highlighted Details

• Broad Format Support: Ingests Markdown, PDF, DOCX, XLSX, PPTX, CSV, EPUB, EML, TXT, VTT, SRT, images (via vision LLM), and numerous code file types.
• Advanced Search & Querying: Features hybrid BM25 + vector search with detailed performance benchmarks, and natural language Q&A with cited answers.
• Ecosystem Integration: Natively integrates with Obsidian vaults and connects to LLM agents via the Message Communication Protocol (MCP), including a wiki_capture tool for AI conversations.
• User Interfaces: Offers a robust CLI, an interactive terminal dashboard (TUI), and an optional embedded Web UI.
• Cost Management: Includes token usage tracking, cost estimation, prompt caching (50-90% savings), and batch API processing (50% cost reduction).

Maintenance & Community

The provided README focuses on technical implementation and lacks explicit details on maintainers, community channels (e.g., Discord, Slack), or project roadmap.

Licensing & Compatibility

• License: MIT License.
• Compatibility: Permissive MIT license allows for broad compatibility with commercial use and integration into closed-source projects.

Limitations & Caveats

The Web UI build process requires Node.js. Image analysis necessitates a vision-capable LLM. Quality metrics indicate areas for improvement, particularly in fact extraction (68.5%), wiki connectivity (60.5%), and cross-reference integrity (50.0%).