Doc OCR

OCR (Optical Character Recognition) for Word documents (.docx) containing scanned pages or image-embedded content. Uses MinerU to extract text from Word file...

Security Scan

The skill's requirements and runtime instructions are consistent with its stated purpose (running the MinerU CLI to OCR .docx files); nothing requested is disproportionate, but be aware the CLI likely sends documents to MinerU's service and the MINERU_TOKEN must be protected.

Purpose & Capability

Name/description (OCR for .docx using MinerU) matches the declared requirements: a mineru-open-api binary and a MINERU_TOKEN. The install options (npm or go install for mineru-open-api) are the expected way to obtain that CLI.

Instruction Scope

SKILL.md only instructs running mineru-open-api on local files or URLs and configuring MINERU_TOKEN. It does not ask the agent to read unrelated files or environment variables. Important caveat: the docs and auth flow imply processing via MinerU's service (token management and API token creation), so document contents may be uploaded to an external service—review privacy requirements before OCRing sensitive documents.

Install Mechanism

Install spec uses npm (mineru-open-api) or go install from a GitHub path — both are reasonable for a CLI. Note that global npm installs run package scripts and that npm packages come from the public registry; if you need higher assurance, inspect the package source or install from the project repo directly.

Credentials

Only MINERU_TOKEN is required and set as the primary credential, which is proportionate for a remote OCR API. Keep the token secret and limit its scope if possible.

Persistence & Privilege

Skill is not always-enabled and does not request system config paths or other skills' credentials. It is user-invocable and can be autonomously called by the agent (normal behavior) but does not request elevated persistence.

Assessment

This skill appears to do what it says: it runs the MinerU CLI to OCR .docx files and requires a MinerU API token. Before installing: (1) confirm you trust the npm package or GitHub repo (inspect source if you need high assurance); (2) treat MINERU_TOKEN like a secret—use a token with minimal scope and do not store it in shared places; (3) assume documents processed may be uploaded to MinerU's servers—do not OCR highly sensitive documents unless you verify local-only processing or run your own MinerU instance; (4) prefer installing from official project releases or from source if you want to audit behavior (npm installs can run scripts).

Use OCR to extract text from Word (.docx) files that contain scanned pages or image-embedded content, using MinerU.

Install

npm install -g mineru-open-api
# or via Go (macOS/Linux):
go install github.com/opendatalab/MinerU-Ecosystem/cli/mineru-open-api@latest

Quick Start

# OCR extraction from .docx (requires token)
mineru-open-api extract report.docx --ocr -o ./out/

# With VLM model for better accuracy on complex image layouts
mineru-open-api extract report.docx --ocr --model vlm -o ./out/

Authentication

Token required:

mineru-open-api auth             # Interactive token setup
export MINERU_TOKEN="your-token" # Or via environment variable

Create token at: https://mineru.net/apiManage/token

Capabilities

• Supported input: .docx (local file or URL)
• OCR is only available via extract (requires token)
• Use --ocr flag to enable OCR on image-embedded content
• Use --model vlm for complex or mixed-content documents
• Language hint with --language (default: ch, use en for English)

Notes

• OCR is NOT available in flash-extract — use extract with --ocr
• If the .docx has a normal text layer, OCR is not needed — use doc-extract instead
• Output goes to stdout by default; use -o <dir> to save to a file or directory
• All progress/status messages go to stderr; document content goes to stdout
• MinerU is open-source by OpenDataLab (Shanghai AI Lab): https://github.com/opendatalab/MinerU