# LlmFiles

> LlmFiles converts documents into Markdown for LLM and RAG pipelines: PDF, DOCX, XLSX, PPTX, EPUB, HTML, CSV, JSON, RTF and 16 more formats. Web app, HTTP API and an MCP server. Free, with no account and no key: 25 conversions per day per IP, 25 MB per file.

Notes for agents:

- To convert one file, POST it to https://llmfiles.com/api/convert as `multipart/form-data` with the field `file`. The JSON response carries the result in `markdown` (plus `filename`, `title`, `size`, `fallback`). This is the same code path as the web UI.
- From an MCP client, connect to https://llmfiles.com/mcp: Streamable HTTP, one JSON-RPC 2.0 message per POST, no SSE stream and no sessions. Tools: `convert_file`, `check_pdf` (per-page verdict before conversion; does not spend quota), `list_formats`. The daily quota is shared with the HTTP API.
- Every guide below also exists as Markdown at the same URL with `.md` appended (`/guide/pdf-to-markdown.md`). Those files are generated from the published page, so they match it. The HTML pages advertise them with `<link rel="alternate" type="text/markdown">`.
- The site loads no third-party scripts, fonts or images, and conversion runs on LlmFiles' own server rather than a third-party converter.

## Product

- [LlmFiles](https://llmfiles.com/): upload a document, get Markdown in the browser. 24 formats, PDF check before conversion, no signup.
- [PDF check](https://llmfiles.com/pdf-check): per-page verdict for a PDF — text layer, scan, encryption, empty pages — so an agent can tell whether OCR is needed before converting.
- [API documentation](https://llmfiles.com/api-docs): `POST /api/convert`, `POST /api/convert/zip`, `POST /api/convert/batch`, `POST /api/pdf-check`, `GET /api/formats`, `GET /api/license`, with request and response examples.

## Guides (each has a `.md` twin)

- [PDF to Markdown](https://llmfiles.com/guide/pdf-to-markdown.md): a PDF stores no headings — what the text layer gives, how headings are inferred, and where the inference fails.
- [DOCX to Markdown](https://llmfiles.com/guide/docx-to-markdown.md): why DOCX converts better than PDF; headings, lists and tables survive, with `|` and line breaks escaped inside cells.
- [XLSX to Markdown](https://llmfiles.com/guide/xlsx-to-markdown.md): what one workbook becomes — merged cells, lost number formats, dates as ISO 8601, formulae as text when no cached value is stored.
- [PPTX to Markdown](https://llmfiles.com/guide/pptx-to-markdown.md): a PPTX is a zip of XML, not a picture; slide text and tables, with cell pipes escaped.
- [EPUB to Markdown](https://llmfiles.com/guide/epub-to-markdown.md): an EPUB is a zip of web pages; what survives conversion and what the empty header row costs.
- [HTML to Markdown](https://llmfiles.com/guide/html-to-markdown.md): the mapping is mechanical; page chrome is dropped before parsing — five cleanup rules with measurements on real pages.
- [CSV to Markdown](https://llmfiles.com/guide/csv-to-markdown.md): values arrive exactly as written, encoding detected on the whole file, `|` and line breaks escaped.
- [JSON to Markdown](https://llmfiles.com/guide/json-to-markdown.md): the top-level value picks the shape; columns are the union of all keys; long tables continue every 1000 rows.
- [Markdown for RAG](https://llmfiles.com/guide/markdown-for-rag.md): what the converter hands a chunker, and why headings are the only reliable boundary in a document.
- [Documents for LLMs](https://llmfiles.com/guide/documents-for-llm.md): what each format hands the model, with token counts per format.
- [PDF to Markdown without the cloud](https://llmfiles.com/guide/pdf-to-markdown-without-cloud.md): three promises that get confused — privacy, offline and on-prem — and how to weigh an upload before deciding.
- [File to Markdown converter](https://llmfiles.com/guide/file-to-markdown-converter.md): twelve files, one table — how twelve formats convert and what each one loses.
- [Batch converting documents to Markdown](https://llmfiles.com/guide/batch-convert-to-markdown.md): three routes for a corpus (API batch, ZIP, CLI) side by side, with numbers from the same corpus.
- [MCP server for PDF to Markdown](https://llmfiles.com/guide/mcp-pdf-to-markdown.md): connecting the hosted MCP server to Claude Code, Claude Desktop, Cursor and VS Code; the three tools, the base64 step and the quota.

## Optional

- [Guide index](https://llmfiles.com/guide): all guides in one page.
- [Pricing](https://llmfiles.com/pricing): what the free tier includes.