PDF to DOCX
Convert PDF files to editable Word (.docx) format using MinerU.
⚠️ Token required.
flash-extractdoes not support DOCX output. You must configure a token viamineru-open-api authbefore using this skill.⚠️ Output to file required. DOCX is a binary format and cannot be streamed to stdout — you must always specify
-o <directory>.
Install
npm install -g mineru-open-api
# or via Go (macOS/Linux):
go install github.com/opendatalab/MinerU-Ecosystem/cli/mineru-open-api@latest
Authentication
Token required — create one at https://mineru.net/apiManage/token:
mineru-open-api auth # Interactive token setup
export MINERU_TOKEN="your-token" # Or via environment variable
Quick Start
# Convert PDF to DOCX (token required, -o is mandatory)
mineru-open-api extract report.pdf -f docx -o ./out/
# From URL
mineru-open-api extract https://example.com/report.pdf -f docx -o ./out/
# With language hint
mineru-open-api extract report.pdf -f docx --language en -o ./out/
# With VLM model for better layout accuracy (complex PDFs)
mineru-open-api extract report.pdf -f docx --model vlm -o ./out/
# Batch convert multiple PDFs
mineru-open-api extract *.pdf -f docx -o ./out/
Capabilities
- Supported input: .pdf (local file or URL)
- Output format: Word (.docx) via
-f docx - Token required (
mineru-open-api authorMINERU_TOKENenv) -o <dir>is mandatory — DOCX cannot stream to stdout- Language hint with
--language(default:ch, useenfor English) - Page range with
--pages(e.g.1-10) - Batch mode supported:
extract *.pdf -f docx -o ./out/
Notes
flash-extractdoes NOT support DOCX output — always useextractwith token- DOCX output cannot be streamed to stdout;
-oflag is required - Use
--model vlmfor PDFs with complex layouts, tables, or mixed content - Use
--model pipelineif you need guaranteed fidelity with no hallucination risk - Output directory will be created if it does not exist
- All progress/status messages go to stderr
- MinerU is open-source by OpenDataLab (Shanghai AI Lab): https://github.com/opendatalab/MinerU