mineru-document-extraction
MinerU document extraction — convert PDFs, scanned documents, images, Word (DOC/DOCX), PowerPoint (PPT/PPTX), and web pages into clean Markd
Open source Repository Open in the app JSON README (API)
About
MinerU document extraction — convert PDFs, scanned documents, images, Word (DOC/DOCX), PowerPoint (PPT/PPTX), and web pages into clean Markdown, HTML, LaTeX, or DOCX. MinerU is an all-in-one CLI tool and agent skill for reliable, high-fidelity document parsing.
Details
- Kind
- Plugins
- Topic
- Files & documents
- Publisher
- decrystal
- Origin
- marketplace
- Category
- ferramentas
- Last push
- 2026-05-08T03:51:52Z
- Repository state
- ativo
- Added
- 2026-08-30 01:48:58
- Updated
- 2026-08-30 01:48:58
- Origin id
decrystal/mineru-claude-plugin/mineru-document-extraction
README
# MinerU MCP Plugin README
This plugin enables Claude to extract high-precision structured data (Markdown, HTML, LaTeX, DOCX) from PDFs, images, and web pages using the MinerU API via the Model Context Protocol (MCP).
## 🚀 Features
- **Document Extraction:** Convert PDFs, images, and Word/PPTx files to structured Markdown.
- **Web Extraction:** Fetch and convert remote web pages via URL.
- **Precision Mode:** High-precision table recognition, formula recognition, and OCR.
- **Flash Mode (Default):** Free, no authentication required, limited to 20 pages / 10 MB. **Used automatically if no API Token is set.**
## 🛠️ Setup
### Important Notice
> **未设置 API Token，将使用 Flash 模式（免费，无需认证，限 20 页 / 10 MB）。如需完整功能，请设置 MINERU_API_TOKEN 环境变量。**
This plugin supports automatic fallback to Flash Mode. If you do not provide a `MINERU_API_TOKEN`, it will default to the free tier.
### Required Environment Variables
To unlock full functionality (Precision Mode), set your API token:
```bash
export MINERU_API_TOKEN="your-token-here"
```
### Obtaining Tokens
1. Visit [https://mineru.net/api-keys](https://mineru.net/api-keys) (official MinerU API Token page).
2. Create a new API token.
3. Set the `MINERU_API_TOKEN` environment variable.
## ⚙️ MCP Configuration
Your MCP config (e.g., `claude_desktop_config.json` or `mcp.json`) should remain as follows. The `env` field will pass the environment variable you set to the MinerU MCP server:
```json
{
"mcpServers": {
"mineru": {
"type": "streamableHttp",
"url": "https://mcp.mineru.net/mcp",
"env": {
"MINERU_API_TOKEN": "your token"
}
}
}
}
```
## 📂 Directory Structure
```text
mineru-claude-plugin/
├── .claude-plugin/
│ └── plugin.json # Plugin metadata
├── .mcp.json # MCP server configuration
├── skills/ # Skill definitions
│ └── mineru-extract/
│ ├── skill.json # Skill metadata
│ └── SKILL.md # Skill instructions
└── README.md # Documentation
```
## 📖 Usage
Once configured, Claude can use the following capabilities through the Mineru MCP server:
### 1. Document Extraction
- **Flash Extract:** Fast conversion for files up to 10 MB and 20 pages.
- **Precision Extract:** High-precision mode requiring an API token, supporting tables and formulas.
### 2. Web Extraction
- **Crawl:** Convert content from a URL to Markdown or HTML.
## ❓ Troubleshooting
- **Authentication Errors:** If you see 401 Unauthorized, verify your token.
- **Flash Mode Fallback:** If you are not using a token, ensure your file is within the 20 pages / 10 MB limit.
- **Feature Limits:** Flash Mode does not support complex table/formula recognition. Use Precision Mode for these features.
## 🔗 Resources
- [MinerU Website](https://mineru.net/)
- [MinerU Extract Skill](https://clawhub.ai/mineru-extract/mineru-document-extractor)
- [Claude Plugins Documentation](https://github.com/anthropics/claude-plugins-official)