mdstill
mdstill is a fast, web-based document to Markdown converter that supports 18 common file formats including PDF, DOCX, PPTX, XLSX, HTML, EPUB, CSV, JSON, XML, RTF, ODT, Apple iWork formats (Pages, Numbers, Keynote), and ZIP archives. It processes files through a layered, layout-aware conversion pipeline that preserves text, tables, headings, and document structure, outputting clean GitHub-flavored Markdown. Most conversions complete in under a second, with no configuration required—users simply upload a file via drag-and-drop, clipboard paste, or file picker, and get a Markdown file ready for immediate use. The tool also offers structured, token-aware output tailored for vector databases and RAG pipelines, making it particularly suited for LLM-centric workflows. Beyond simple conversion, mdstill provides a range of output options: users can copy the Markdown to clipboard, download an .md file, or pipe the result directly into a vector database with minimal post-processing. A reverse conversion from Markdown back to formats like PDF, Word, HTML, EPUB, PowerPoint, RTF, ODT, and LaTeX is also available. The service operates entirely in-browser for basic use without storing uploads, and a free tier allows file sizes up to 10 MB; signing in unlocks larger limits and additional features. mdstill is designed explicitly for AI and developer ecosystems. It ensures that LLMs like ChatGPT, Claude, and Gemini receive well-structured input that preserves headings, tables, and code blocks, which reduces token waste and improves comprehension. For retrieval-augmented generation (RAG), the Markdown output retains semantic structure that aids chunking and retrieval accuracy when used with frameworks such as LangChain and LlamaIndex. Other use cases include importing documents into knowledge bases like Obsidian and Notion, feeding specs into IDEs (Cursor, Copilot), and automating conversion pipelines via n8n, Zapier, or a REST API. The API accepts multipart/form-data and returns Markd
- Category
- LLM
- Pricing
- Free
Problem solved
Document-heavy workflows in AI, software development, and content management often require converting files from proprietary formats (PDF, DOCX, PPTX, etc.) into plain-text Markdown to be ingestible by LLMs, used in RAG pipelines, or versioned in repositories. Manual conversion is time-consuming, error-prone, and frequently destroys tables, headings, code blocks, and other structural elements. mdstill solves this by providing an instant, server-side conversion service that reliably preserves document structure and formatting, delivering clean Markdown that is immediately usable for cutting-and-pasting into LLM prompts, indexing in vector databases, importing into knowledge tools, or piping through automation workflows.
Use cases
- Converting PDF reports or Word documents into Markdown for use as context in ChatGPT or Claude prompts without losing tables, headings, or code blocks.
- Preparing documents for Gemini by converting formats like PDF to Markdown to maintain clean heading hierarchy and code fences.
- Feeding converted documents into RAG pipelines built with LangChain or LlamaIndex, where Markdown structure improves chunking and retrieval accuracy.
- Importing legacy Word, PDF, or Keynote files into Obsidian, Notion, or Logseq vaults as well-formatted Markdown.
- Transforming customer-provided DOCX/PDF specifications into Confluence pages or Jira tickets with intact structure.
- Automating document-to-Markdown conversion in no-code workflows (n8n, Zapier, Make) via the REST API, then routing the output to LLM nodes or databases.
Audience
- Developers
- ML Engineers
- Data Scientists
- Content Creators
Pros
- Extremely fast conversion (most files under a second) with no configuration needed.
- Broad format support covering 18 common document types including Apple iWork and ZIP.
- Preserves complex structure—tables, headings, code blocks—yielding clean GitHub-flavored Markdown.
- Free tier available, with optional sign-in for higher limits; no files stored on server.
- Offers both web interface and API for integration into automated pipelines (n8n, Zapier, etc.).
Cons
- Free tier limited to 10 MB file size; larger files require sign-in and potentially a paid plan.
- No self-hosted option; reliance on external service may be a concern for sensitive documents.
- May encounter issues with extremely complex or malformed documents (behavior not detailed for edge cases).
Official website: https://mdstill.com/
Canonical LambdaBase page: https://www.lambdabase.com/tools/mdstill