Sage-wiki

sage-wiki ingests documents, notes, code, email, and other supported formats, then uses an LLM compilation pipeline to produce interconnected Markdown articles, summaries, concepts, and a knowledge graph. Users can search the resulting corpus with hybrid lexical, vector, and graph retrieval, ask cited questions, inspect provenance, and capture new knowledge.

The product supports humans and AI agents through a web UI, terminal dashboard, MCP tools, REST API, and Obsidian-compatible Markdown. Its graph features include typed relations, evidence, confidence, entity resolution, contradiction handling, and multi-hop graph queries. It runs as a Go binary, with local-model support, SQLite or PostgreSQL/pgvector storage, Docker Compose deployment, shared servers, Git synchronization, and federated hubs.