# Molio [Health: Active]

**Category:** 🧠 Knowledge & Memory  
**Repository:** https://github.com/zhuzhaoyun/Molio  
**GitHub Stars:** 441  
**npm Downloads (last month):** 511  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/molio

## Description
A local-first personal knowledge layer for AI agents. Build evolving knowledge spaces with LLM Wiki, knowledge graphs, and agent workflows.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "molio": {
    "command": "npx",
    "args": ["-y","@molio/cli"]
  }
}
```

## Documentation

## What Molio does

Molio provides a local-first knowledge layer for AI agents. It organizes accumulated material into knowledge spaces that agents can use for research, writing, question answering, and analysis. The underlying content remains on the user’s machine, with Markdown serving as the durable storage format.

The system is designed for more than simple document lookup. Its Wiki engine extracts entities and concepts from the Markdown layer, then creates cross-links and layered indexes. A knowledge graph presents relationships between items visually and becomes richer as new material and task results are added.

Molio can also work with ready-made knowledge bases covering subjects such as literature, history, philosophy, medicine, traditional Chinese medicine, and AI engineering. These can be imported for immediate exploration instead of being assembled from scratch.

## How it works

Content enters through several paths. Plain Markdown, text, HTML, CSV, and other text files can be read directly. PDFs and Office files are processed with layout analysis, OCR, and table reconstruction; images are converted through OCR. Web pages can be clipped or extracted, and an Obsidian vault can be opened in place with its original files kept read-only.

Large books use chunked preprocessing and a digest pipeline with resumable progress. After processing, the resulting Markdown becomes the shared layer used by the Wiki engine and agents. Agents can work inside that knowledge space through supported agent environments, and their outputs are deposited as Markdown for later tasks.

A single Molio installation can also serve multiple channels. The Web Console supports text, images, files, and voice. WeChat and Feishu support text, images, and files, while some additional channel capabilities are marked as planned.

## Tools and capabilities

- Expose a local personal knowledge space to AI agents through MCP.
- Import and normalize text, PDFs, Office documents, images, web pages, and large books.
- Generate linked wiki content, indexes, and a visual knowledge graph from Markdown.
- Let agents research, write, answer questions, and analyze accumulated material.
- Write agent results back into the knowledge base as Markdown.
- Open an existing Obsidian vault without modifying its original files.
- Use ready-made subject knowledge bases for areas including history, literature, philosophy, medicine, and AI engineering.
- Publish prepared content through doocs/md to more than 30 platforms.

## Limitations and notes

PDF and Office conversion uses the bundled docling skill. The first conversion downloads the tool and approximately 500 MB of models; later conversions reuse the local cache. OCR and layout processing therefore add local setup and storage requirements beyond the Markdown workflow.

The channel table distinguishes available functions from planned ones. Telegram, Slack, and Discord capabilities are listed as planned, and voice or group features are not uniformly available across WeChat and Feishu. The knowledge workflow is centered on local files and Markdown, so teams seeking a hosted, shared repository should evaluate that storage model carefully.

Agent access is presented for Claude Code, Codex, Gemini CLI, and Qwen Code, while the Web Console provides a unified interface with streaming output. A phone-based chat path is also available through WeChat onboarding by QR code.

_Full upstream README: https://allmcps.com/mcp/molio/readme_