LLMtoMD logoLLMtoMD
All docs

How-to guides

Convert a web page or URL

Turn any article, docs page, or file URL into clean, AI-ready Markdown your agent can read and store.

The fastest way to add something to your library is by URL. If it's reachable on the web — an article, a documentation page, a hosted PDF or Office file — your agent can convert it with one request.

Convert a single page

Convert https://example.com/blog/scaling-postgres to Markdown and save it.

Behind the scenes your agent calls convert_url, waits for the job to finish, and gets back clean Markdown plus a document_id. The page is now in your library and searchable.

Convert a hosted file

The same tool handles file URLs — PDFs, Word/Excel/PowerPoint, and images:

Convert https://example.com/reports/q3.pdf and store it in my library.

For scanned PDFs or images, add an OCR language hint so text is extracted accurately:

Convert this scanned URL with OCR for English and German.

(That maps to ocr_langs: "eng+deu".)

Convert an entire site

Set crawl to pull a whole documentation site in one go. It follows same-origin links, bounded so it won't wander off-domain:

Crawl https://docs.example.com and add the whole site to a "Vendor API" project.

This is the quickest way to give your agent a vendor's full docs as searchable memory.

What you get back

  • Clean Markdown with headings, lists, tables, and code preserved.
  • A document_id your agent reuses for get_document, search_documents, and ask_documents.
  • The document stored permanently in your library until you delete it.

When a conversion is slow

Large or slow sources may not finish within the wait window. The agent gets a "still processing" note; the document shows up in list_documents a little later — just ask "is it ready yet?" and the agent will check.

Related