Convert mixed documents into agent-ready Markdown and JSON with DocStrange
Convert PDFs, scans, office files, images, and URLs into clean Markdown, structured JSON, CSV, or HTML before an agent ingests or reviews the content.
npx skills add agentskillexchange/skills --skill convert-mixed-documents-into-agent-ready-markdown-and-json-with-docstrange
Use DocStrange when an agent workflow needs reliable document preparation before retrieval, review, extraction, or downstream automation. The operator provides PDFs, scanned images, office documents, spreadsheets, or URLs, then converts them into clean Markdown, structured JSON, CSV, or HTML with OCR, table handling, and optional schema-guided field extraction. Invoke this instead of using the hosted product normally when the job is a repeatable ingestion or review pipeline where outputs need to be handed to an agent, stored in a RAG index, or checked in a local/private processing mode. The scope boundary is document-to-agent-data conversion; it is not a generic OCR platform, document management system, or broad SDK listing.
What this skill actually does
Inputs and prerequisites: Python 3.8+, docstrange package, optional CUDA GPU for local processing, optional MCP-compatible client or local web UI.
Setup notes: Install with pip install docstrange for library use. For the local web UI, install with pip install “docstrange[web]” and run docstrange web or python -m docstrange.web_app . Choose cloud mode for managed processing or GPU mode for local/private document conversion.
Source and verification boundary: use https://docstrange.nanonets.com/ as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.