skills.rest
DawnMoon1542
DawnMoon1542COMMUNITY
Explore all skills by this author →
GitHub

mineru-extract

Convert URLs and documents into Markdown using the MinerU parsing API.

7|Updated May 13, 2026
Search Tags:#ocr#markdown#document conversion#web scraping#pdf extraction#mineru#url parsing
One-click install
npx skills add https://github.com/DawnMoon1542/agents-skills --skill mineru-extract-dawnmoon1542
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mineru-extract
Source: https://github.com/DawnMoon1542/agents-skills/tree/main/search-skills/mineru-extract
Command: npx skills add https://github.com/DawnMoon1542/agents-skills --skill mineru-extract-dawnmoon1542

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Standard web fetching often fails on protected pages or produces messy output for complex documents. This Skill uses the official MinerU API to convert URLs (HTML pages like WeChat articles, or direct PDF/Office/image links) into clean Markdown with high-fidelity layout, table, formula, and OCR parsing. ## Core Features & Use Cases - URL-to-Markdown Conversion: Submit any URL to MinerU, poll for completion, and download the extracted Markdown plus structured JSON outputs. - MCP-Style Batch Interface: Parse multiple sources at once via mineru_parse_documents.py, which returns a stable JSON contract with per-item results and errors. - Model Selection & OCR: Automatically picks pipeline for documents/images and MinerU-HTML for web pages, with options for OCR, table, and formula recognition. - Use Case: You need to archive a WeChat article that web_fetch renders poorly. Run the parse script with the article URL and receive clean Markdown plus the full result zip cached locally. ## Quick Start Set your MINERU_TOKEN in the skill's .env file, then ask the AI to parse a URL into Markdown using the mineru_parse_documents script with the MinerU-HTML model.

Frequently Asked Questions about mineru-extract

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I convert a URL to Markdown with the MinerU API?▼

Run the mineru_parse_documents.py script with --file-sources set to your URL and --model-version MinerU-HTML. The script submits the task, polls until done, downloads the result zip, and returns a JSON summary with the markdown path.

How do I extract text from PDF or image URLs with OCR?▼

Pass the PDF or image URL to the script with the pipeline model and the --enable-ocr flag. The scripts auto-select pipeline for URLs ending in .pdf, .doc, .ppt, .png, or .jpg, and OCR applies to pipeline and vlm models.

What authentication does the MinerU API require?▼

The MinerU API requires a Bearer token from mineru.net set as the MINERU_TOKEN environment variable. You can place it in a .env file at the skill root, which the scripts load automatically.

Can I parse local files with this MinerU workflow?▼

Local file paths are not supported in the current workflow; only public URLs are accepted. The scripts return an error suggesting you provide an accessible URL or add MinerU batch upload support for local files.

Why does MinerU fail to fetch some URLs?▼

MinerU may fail on pages with anti-bot protection, geo-restrictions, or login requirements. The skill recommends falling back to an HTML file, PDF, or long screenshot via MinerU's batch upload endpoints, and reporting the failing URL with its err_msg.

Related Skills

Explore More Content & Communication Skills→
aaronjmarsaaronjmars
17

RSS Digest

Fetch RSS/Atom feeds from a YAML file and summarize recent entries.

Community
Intermediate
kreuzberg-devkreuzberg-dev
9.2k

chunking

Split extracted text into sized chunks for LLM context windows and RAG ingestion.

Official
Intermediate
zjp1997720zjp1997720
12

web-clipper

Clip web articles into local Markdown with frontmatter and original URLs.

Community
Intermediate
opendatalabopendatalab
185

MinerU Document Extractor

Convert PDFs, images, and Office files into Markdown, HTML, or JSON.

Official
Intermediate