Use the official MinerU (mineru.net) parsing API to convert a URL (HTML pages like WeChat articles, or direct PDF/Office/image links) into clean Markdown + structured outputs. Use when web_fetch/browser can’t access or extracts messy content, and you want higher-fidelity parsing (layout/table/formula/OCR).
Use MinerU as an upstream “content normalizer”: submit a URL to MinerU, poll for completion, download the result zip, and extract the main Markdown.
We align to the MinerU MCP mental model, but we do not run an MCP server.
scripts/mineru_parse_documents.py
--file-sources (comma/newline-separated){ ok, items, errors }scripts/mineru_extract.pyAuth:
MINERU_TOKEN (Bearer token from mineru.net)Default model heuristic:
.pdf/.doc/.ppt/.png/.jpg → pipelineMinerU-HTML (best for HTML pages like WeChat articles)Put secrets in skill root .env (do not paste into chat outputs):
# In the mineru-extract skill directory: .env
MINERU_TOKEN=your_token_here
MINERU_API_BASE=https://mineru.net
MCP-style wrapper (returns JSON, optionally includes markdown text):
python3 mineru-extract/scripts/mineru_parse_documents.py \
--file-sources "<URL1>\n<URL2>" \
--language ch \
--enable-ocr \
--model-version MinerU-HTML
If you want the markdown content inline in the JSON (can be large):
python3 mineru-extract/scripts/mineru_parse_documents.py \
--file-sources "<URL>" \
--model-version MinerU-HTML \
--emit-markdown --max-chars 20000
Low-level (single URL, print markdown to stdout):
python3 mineru-extract/scripts/mineru_extract.py "<URL>" --model MinerU-HTML --print > /tmp/out.md
The script always downloads + extracts the MinerU result zip to:
~/.openclaw/workspace/mineru/<task_id>/
It writes:
result.zipIt prints a JSON summary to stderr with paths:
task_id, full_zip_url, out_dir, markdown_path--model: pipeline | vlm | MinerU-HTML (HTML requires MinerU-HTML)--ocr/--no-ocr: enable OCR (effective for pipeline/vlm)--table/--no-table: table recognition--formula/--no-formula: formula recognition--language ch|en|...--page-ranges "2,4-6" (non-HTML)--timeout 600 / --poll-interval 2err_msg and keep an original-source link in outputs.