iwe-org/iwe

What it solves

IWE solves the problem of managing personal knowledge in a way that is both human-readable and AI-accessible. It prevents vendor lock-in by using plain markdown files while providing the structured, graph-like connectivity usually reserved for proprietary databases, allowing users to maintain a "second brain" that AI agents can efficiently query and update.

How it works

IWE transforms a local directory of markdown files into a knowledge graph using two types of links: "inclusion links" (nesting a topic under another) and "cross-references" (inline links). It uses a Rust-based engine to process these files rapidly. To bridge the gap between humans and AI, it provides a Language Server Protocol (LSP) for IDEs (like VS Code and Neovim) and a Model Context Protocol (MCP) server and CLI for AI agents, allowing them to search, retrieve, and refactor notes with full hierarchical context.

Who it’s for

It is designed for individuals who want to maintain a local-first knowledge base in markdown and developers who want to give their AI agents (such as Claude or Cursor) structured, queryable access to their personal notes and documentation.

Highlights

• Local-first ownership: Uses plain .md files and git for versioning with no cloud or database requirements.
• AI-native integration: Implements the Model Context Protocol (MCP) for seamless connection with AI tools.
• IDE-grade features: Provides LSP support for autocomplete, refactoring, and navigation across editors.
• High performance: Built in Rust, capable of processing 20,000 files in under a second.
• Flexible hierarchy: Supports multiple parents, allowing a single note to belong to multiple topics without duplication.