zhuzhaoyun/Molio

What it solves

Molio creates a local-first personal knowledge base that AI agents can read and interact with. It solves the problem of fragmented personal data (notes, PDFs, books, chat logs) that typically cannot be read by AI, allowing users to build a growing, structured knowledge universe on their own machine without relying on third-party servers.

How it works

Molio processes diverse inputs—including Markdown, PDFs, Word, PPT, Excel, images, and Obsidian vaults—into a unified Markdown layer using tools like docling for OCR and layout analysis. A Wiki engine then extracts entities and concepts to create cross-links and layered indexes. AI agents (such as Claude Code, Codex, Gemini CLI, and Qwen Code) operate within this space to research, write, and analyze data. The outputs of these tasks are written back into the knowledge base as Markdown, allowing the system to evolve over time.

Who it’s for

It is designed for individuals who want to maintain a private, local-first AI knowledge base, researchers, and those who whom want to use pre-made structured knowledge graphs (e.g., in history, philosophy, or medicine) to power their AI interactions.

Highlights

• Local-first privacy: All data stays on the user's machine.
• Broad input support: Handles everything from scanned PDFs and million-word books to Obsidian vaults.
• Agent orchestration: Integrates with multiple AI runtime CLIs for streaming responses.
• Self-growing base: Task outputs are saved back as reusable assets.
• Multi-channel access: Supports Web Console, WeChat, and Feishu.
• Ready-made graphs: Provides pre-organized knowledge bases for specific professional domains.