Merge every PDF in ~/Documents/receipts/august into one file called august-receipts.pdf.
Paste this into Claude with the server connected.
pdf_merge. Free merges up to 5 files at a time, which for a month of receipts usually
means merging in batches and then merging the results, and that works because the output of a merge is
just another PDF. Pro removes the count.
How many pages is august-receipts.pdf and what is on the first one?
Paste this into Claude with the server connected.
pdf_count and pdf_text, both free and unmetered.
pdf_text pulls the text layer, which on a receipt printed to PDF is the whole receipt and
on a phone photo saved as PDF is nothing at all. If it comes back empty, that page is an image and no
text tool in this repository will read it: there is no OCR here, and there is not going to be a claim
that there is.
Stamp august-receipts.pdf as PAID.
Paste this into Claude with the server connected.
pdf_stamp with PAID or DRAFT is free. Custom stamp text and colours, and the business
watermark, are Pro. Edits on the free tier work on files up to 30 pages.
pdf_split takes one file apart, pdf_pages pulls out a range, and
pdf_rotate fixes the scan that came in sideways. Page reordering is Pro. The usual reason
to split is that a scanner produced one 40-page file containing eleven separate documents, and each
of them needs to go somewhere different.
The PDF server is pure JavaScript with no network call at all. Receipts frequently carry a home address, a card's last four digits and a client name, and an online merge tool sees all of it. This one opens the paths you named and writes the file you asked for.
Zip the merged PDF and the CSV exports into accounts-august.zip.
Paste this into Claude with the server connected.
The zip server packs a folder with a glob. Free is 20 archives a calendar month at up to 25 MB and
200 entries, and reading an archive is never metered on any tier. All figures on this page come from
data/facts.json; there are 10 tools on the PDF server, counted from
data/tools.json.
Only if the scan carries a text layer, which many scanners add and phone cameras do not. pdf_text returns what is there and nothing when there is nothing. There is no OCR in this repository.
No. The pages are copied as they are, in the order you named the files. Nothing is recompressed and no quality is lost, which also means the merged file is roughly the sum of the inputs in size.
It applies to edits: stamping, rotating, splitting. Info, page count, text extraction and merging up to 5 files are not affected by page count. The distinction is that reading a document is not metered and rewriting one is.
The image server resizes, converts, compresses, crops and strips metadata, up to 4 megapixels and batches of 5 on free. Stripping metadata is the one to know about for receipts: a phone photo carries GPS coordinates until something removes them.
All MCP servers and prices · All guides · Buy the bundle $39