xoai/sage-wiki

sage-wiki is a graph memory and knowledge base that AI agents and humans build and query together. Drop in documents; an LLM compiler turns them into an interlinked wiki with a knowledge graph. One Go binary scales it from a personal vault to a team hub to a company knowledge graph.

What it solves

sage-wiki is a graph-based memory and knowledge base designed for collaborative use between AI agents and humans. It solves the problem of fragmented information by transforming raw documents (PDFs, emails, code, etc.) into an interlinked, structured wiki with an underlying knowledge graph. This allows for more precise retrieval than standard vector search, enabling AI agents to answer multi-hop relational questions with evidence and citations.

How it works

The system uses an LLM-powered "compiler" pipeline that processes source documents into summaries and interconnected markdown articles. It employs a hybrid search mechanism combining lexical (BM25), vector, and graph proximity to retrieve information.

Key technical features include:

  • Evidenced Graph: An optional pass that extracts typed entities and relations with confidence scores and source provenance.
  • Entity Resolution: Merges surface-form variants (e.g., "K8s" and "Kubernetes") into single nodes.
  • MCP Integration: Exposes 19 tools via the Model Context Protocol (MCP), allowing AI agents to search, capture, and query the graph directly.
  • Tiered Compilation: Optimizes LLM budget and speed by indexing large volumes of documents (up to 100K+) efficiently.

Who it’s for

  • Individuals: Users with existing Obsidian vaults who want an AI-enhanced personal knowledge base.
  • Teams: Groups sharing a wiki via Git or a self-hosted server to collaborate on a shared knowledge graph.
  • Company-wide: Organizations needing a scalable knowledge base backed by PostgreSQL/pgvector for large-scale ingestion.

Highlights

  • Hybrid Retrieval: Fuses vector search with graph traversal for complex relational queries.
  • Agent-Human Synergy: Provides a TUI, Web UI, and MCP tools for agents to interact with the same data.
  • Provenanced Answers: Every answer is grounded in serialized graph edges with citations tracing back to the source document.
  • Broad Format Support: Automatically extracts text from PDFs, Word, Excel, PowerPoint, EPUB, emails, and images (via vision LLMs).
  • Obsidian-native: Compiled output is stored as plain markdown, making it compatible with Obsidian.

Related

  • Project
    nashsu/llm_wiki

    A desktop application that automatically turns documents into a structured, interlinked personal wiki using LLMs, featuring a knowledge graph and automated deep research.

  • Project
    VectifyAI/OpenKB

    An open-source system that compiles raw documents into a structured, interlinked wiki-style knowledge base using LLMs and vectorless, reasoning-based retrieval.

  • Project
    atomicstrata/llm-wiki-compiler

    llm‑wiki‑compiler is a Node/TS tool that compiles raw documents into a citation‑aware, typed markdown wiki, with review gates, lifecycle profiles, an SDK, and an MCP server for LLM agents.

  • Project
    sdyckjq-lab/llm-wiki-skill

    A system for building a structured, interconnected knowledge base from fragmented data, featuring an interactive knowledge graph and integration with AI CLI agents.

  • Project
    GD4AI/obsidian-llm-wiki

    An Obsidian plugin that transforms your notes into a structured, queryable wiki using graph-based retrieval instead of vector databases.