Summary
Length caps on every conversion, which is what keeps a long PDF usable.
Each tool takes a maximum output length, and the directory tool takes a file limit too — so pointing it at a folder of long reports fails politely instead of consuming everything. Conversion happens in memory with no temporary files left behind.
What it is
A small document-conversion server. It reads files from disk or fetches a URL, converts the content to Markdown in memory, and caps the output length so a large document does not swamp the conversation.
What you get
- convert_file — one local file to Markdown, with a maximum output length you can set
- convert_url — fetch a URL and convert what comes back
- batch_convert — several files in one call, with a per-file length cap
- convert_directory — every supported file in a directory, optionally recursive, with a file limit
- extract_metadata — metadata from a file without converting the whole thing
- list_supported_formats — what it will accept
- Support for PDF, DOCX, HTML, EPUB, CSV, JSON, images via OCR and plain text
Requirements
Python, or the bundled Dockerfile. Install the requirements and run the module over stdio; with Docker, mount the directory holding the files you want to convert. No credentials.
Setup effort
Build from source — clone the repository and build it, then point your client at the binary