lucasastorian/llmwiki

What it solves

LLM Wiki creates an autonomous, self-maintaining "second brain" or personal Wikipedia. It solves the problem of scattered research, notes, and web clippings by automatically synthesizing these fragmented sources into a structured, persistent knowledge base, preventing institutional or personal memory loss over time.

How it works

Users capture information via a Chrome extension (for web pages and PDFs) or by uploading files (Markdown, Office docs, etc.) into a workspace. The system uses a Model Context Protocol (MCP) server to connect an LLM (like Claude) to the data. A scheduled "Claude Routine" then runs autonomously—typically nightly—to read new sources, update existing wiki pages, create new ones, and maintain cross-references and citations without manual intervention.

Who it’s for

• Individuals wanting a personal knowledge base of everything they have read and thought about.
• AI Agents that need a personalized context layer to apply a user's specific mental models.
• Organizations seeking to build a self-maintaining institutional memory layer to capture know-how that usually resides only in employees' heads.

Highlights

• Autonomous Maintenance: Uses scheduled routines to synthesize and update the wiki automatically.
• MCP Integration: Connects directly to Claude.ai, Claude Code, and other MCP-compatible apps.
• Multi-format Support: Handles PDFs, Word, PowerPoint, Excel, and web pages (with OCR support via Mistral).
• Flexible Deployment: Supports both a local mode (SQLite + filesystem) and a hosted mode (Postgres + S3).
• Knowledge Visualization: Includes a graph viewer to visualize relationships between concepts and entities.