What file size and page limits does Mineru have?
Single file: 200 MB max and 600 pages max. Batch: up to 200 files per request.
@linxule
基本信息
暂无概览
使用下面的配置,将此服务器添加到你的 MCP 客户端。
{
"mcpServers": {
"mineru": {
"command": "npx",
"args": [
"-y",
"mineru-mcp"
],
"env": {
"MINERU_API_KEY": "your-api-key-from-mineru.net"
}
}
}
}Parse a document URL
Check task progress, get download URL
Parse multiple URLs (max 200)
Get batch results with pagination
Upload local files for batch parsing
Download results as named markdown files
Mineru is an MCP (Model Context Protocol) server for the MinerU document parsing API. It extracts text, tables, and formulas from PDFs, DOCs, and images, and is designed for use with AI coding assistants like Claude Code, Codex CLI, and Gemini CLI.
Install via npx -y mineru-mcp (requires Node.js 18+ and a MinerU API key). Configure the server in your MCP client (e.g., Claude Desktop, VS Code, Cursor, Windsurf) by setting the command to npx, args to -y mineru-mcp, and adding the environment variable MINERU_API_KEY. Then use the provided tools: mineru_parse, mineru_status, mineru_batch, mineru_batch_status, mineru_upload_batch, and mineru_download_results.
Single file: 200 MB max and 600 pages max. Batch: up to 200 files per request.
The daily quota is 2000 pages at high priority.
Mineru offers two models: `pipeline` (default) for fast processing of simple documents, and `vlm` for complex documents requiring 90%+ accuracy.
Yes, a MinerU API key is required. Get one at [mineru.net](https://mineru.net). Set it as the `MINERU_API_KEY` environment variable.
No, Mineru uses local stdio transport. ChatGPT only supports remote MCP servers over HTTPS, so it is not directly compatible.

Model Context Protocol Servers
不止于 CSS
Open-source MCP to use Blender with any LLM

Connect your AI assistant to your own YOUB endurance training account. Read recorded activities with full analysis, planned sessions, race goals, wellness (HRV, sleep, resting HR) and your athletic profile. Create or mov

B2B competitive intelligence, live sales battlecards, 5-axis ICP prospect fit scoring, and tech stack teardowns over MCP and REST with cited public-web evidence.
评论