lucasastorian/llmwiki

What it solves

LLM Wiki addresses the problem of scattered research, reading materials, and fragmented notes that are often forgotten over time. It transforms these disparate sources into a persistent, self-maintaining "second brain" or personal Wikipedia, preventing institutional or personal memory loss by autonomously synthesizing information into a structured knowledge base.

How it works

The system uses a combination of a Chrome extension for clipping web content and a local or hosted workspace for uploading documents (PDFs, Office files, Markdown, etc.). An AI agent (specifically Claude) connects to this workspace via the Model Context Protocol (MCP), allowing it to read, search, and write to the wiki. Users can set up "Claude Routines"—scheduled prompts—that instruct the AI to nightly scan for new sources, highlights, and notes, and then autonomously update wiki pages, create new entries, and manage cross-references and citations.

Who it’s for

• Individuals wanting a personal, automated knowledge base of their reading and research.
• AI Users who want to provide their LLMs with a context layer based on their own mental models and specific knowledge.
• Organizations seeking to build a self-maintaining institutional memory layer to capture know-how that typically resides only in employees' heads.

Highlights

• Autonomous Maintenance: Uses scheduled AI routines to synthesize new information without manual updating.
• MCP Integration: Seamlessly connects with Claude.ai, Claude Code, and other MCP-compatible applications.
• Clipper Extension: A Chrome extension that captures web pages and PDFs along with user highlights and margin notes.
• Flexible Storage: Supports both local mode (SQLite and filesystem) and hosted mode (Postgres and S3).
• Rich Visualizations: Supports native cross-linking, a graph viewer for concept relationships, and Mermaid diagrams/SVGs.