Agnos vs rag-scrubber: when to use which

If you write Python and already pull the text out of your PDFs page by page, rag-scrubber is the better pick, more so if you want it free or work on a Mac or Linux: a few lines of code take repeated headers and footers, and page numbers such as “Page 3”, out of that text. If you want whole folders of text-based PDFs, Office files and web pages cleaned on Windows without code, with more kinds of clutter removed and a list of what was removed and the rule that removed it, Agnos is made for that job.

Gidion makes Agnos and wrote this page, so read it with that in mind. Every statement about rag-scrubber comes from its PyPI entry and its GitHub repository (its README, code and tests), checked on 17 October 2026, and describes rag-scrubber 0.1.4, released on 5 January 2026. Each source is listed under Sources.

At a glance

Compared	Agnos	rag-scrubber
What it is	A Windows app, made by Gidion, that cleans documents with named rules before they go into AI tools, and lists what it removed.	A small, free Python library by Mugesh Kumar M. It removes headers and footers that repeat across the pages of text taken from a PDF, and page numbers written as “Page 3” or “3 of 10”. It also joins words split by a hyphen at the end of a line.
Who it’s for	People preparing documents for Custom GPTs, NotebookLM, Claude Projects, Copilot, a RAG index or training data, without writing code.	Developers building RAG pipelines in Python who already take the text out of each page with a PDF library. Its README names PyPDF2 and PDFMiner.
Where it runs	Your PC. Documents are never uploaded; Agnos goes online only to activate, check or deactivate a licence.	Inside your own Python code, on your computer or a server. Its code uses only Python’s standard library and makes no network calls.
Needs code	No. It installs from the Microsoft Store, with no Python to set up. Its command line is coming to the Store version in an update.	Yes. Python 3.7 or newer and `pip install rag-scrubber`, then a few lines of Python. There is no command line or app.
File types	Text-based PDF, Word (`.docx`), PowerPoint (`.pptx`), Excel (`.xlsx`), web pages, CSV, JSON, plain text, Markdown and source code. Old Office formats, OpenDocument, images, audio, video and archives are refused.	It opens no files. You give it a list of strings, one for each page, already taken out of a PDF by another library.
Headers and footers	In the RAG / Retrieval Corpus purpose, it looks at the first two and last two lines of each PDF page or slide. There it removes page markers such as “Page 3 of 12” and “- 3 -”, bare page numbers that count up with the pages, and a short line that repeats there on at least three pages, even when its numbers change. In a PDF or deck with two or more pages, a line in the body of a page is never touched. In a one-page PDF or a one-slide deck, a line that is only a page marker or a “Confidential” banner is removed wherever it is.	It looks at the first three and last three lines of each page. A line of more than three characters that appears there, exactly the same, on at least 40% of the pages (a threshold you can change) is removed wherever it appears; this needs at least two pages. By default it also removes any line, anywhere on a page, that starts with “Page” and a number, or with “Copyright” or “All rights reserved”, or that is only “Confidential”, “3 of 10” or a web address starting with http:// or https://. A page number on its own, such as “12”, is never removed, and one such as “- 12 -” or “12 / 40” stays at the default threshold unless the file has only two pages.
Scanned PDFs (OCR)	No OCR. A page with no text layer is skipped and named in the file’s warnings; a PDF with no text at all is refused.	No OCR, and it doesn’t read PDFs itself. It cleans whatever page text you give it, so it can also clean text from an OCR engine you run first.
Tables	Read from Word, PowerPoint and web pages, and the RAG / Retrieval Corpus purpose keeps them as written. In PDFs, only with the slower Layout reading mode. Excel sheets become tab-separated rows, without formulas or formatting.	It doesn’t read tables. They reach it as the lines of text your PDF library wrote, and stay that way apart from the lines it removes and the words it joins.
What it outputs	A RAG corpus in JSONL (each piece with an id, its source file and its pages, slides, sheet or headings), cleaned `.txt` files, training data with a validation split, or all of these with a record of what was removed.	One string for the whole document, with a blank line between pages. Page numbers are not recorded.
Shows what it removed, and why	Yes. Review names the rule that removed each piece, listing up to 500 removals in a file, and the Everything export adds `rejected.jsonl`, with every removed piece and its reason.	No. It returns only the cleaned text.
Price	A 14-day trial of everything in Pro, no card. Then $99 once for one person on up to 2 devices, or $399 once for up to five people at one company, on up to 10 devices. Tax is added where it applies. No subscription.	Free.
Operating systems	64-bit Windows 10 (version 2004 or later) and Windows 11.	Any system with Python 3.7 or newer. It is pure Python with no dependencies, published as one package for every platform.
Licence	Commercial, under an end user licence agreement.	Its README says “MIT License. Free to use for commercial and personal projects.” The repository has no LICENSE file, and its PyPI entry names no licence.

Where rag-scrubber is the better choice

You are building a Python pipeline. rag-scrubber is a library you call from your own code: you pass it the text of each page and get the cleaned text back. It fits in a script, a notebook or a service, after whichever PDF library you already use, and adds no dependencies. Its tests import it as from rag_scrubber.cleaner import RAGScrubber. Agnos is a desktop app, and its command line is not in the Microsoft Store version yet.
Your text comes from somewhere Agnos can’t read. Because it cleans text rather than files, rag-scrubber can tidy pages that come from an OCR engine, or from any extractor you choose. Agnos has no OCR, so a scanned page gives it nothing to read, and it reads only the file types in Supported files.
You want to tune it. One number, threshold, sets the share of pages a line must appear on to count as a header or footer; its README calls 0.1 very aggressive and 0.9 very safe. A second option in its code, use_regex_fallback, switches off its fixed patterns for page numbers, copyright lines and the rest. In Agnos, a running header has to repeat at the edge of at least three pages, and that number is not a setting.
You don’t use Windows. It is pure Python with no dependencies, for Python 3.7 or newer, so it runs wherever Python does. Agnos runs only on 64-bit Windows 10 and 11.
You want it free, and small enough to read. It costs nothing, and the whole library is one file of under 100 lines that you can read in a few minutes. Its README says it is under the MIT License, but the repository has no LICENSE file, so check that this is enough for your company before you build on it.

Where Agnos is the better choice

You don’t write code. Agnos installs from the Microsoft Store and brings everything it needs: no Python, no PDF library to choose, no script to keep. You drop in a folder, check the result and press export. rag-scrubber needs Python, a PDF library to take the text out of each page first, and your own code to save what it returns.
You have a folder of mixed files. One project takes text-based PDFs, Word, PowerPoint and Excel files, saved web pages, CSV, JSON, Markdown and plain text together, and cleans each one as it arrives. rag-scrubber opens no files; it cleans the page text you pass it.
You need more than headers and footers taken out. Besides page numbers and running headers and footers at the edges of PDF pages and slides, Agnos’s rules remove navigation menus, cookie banners, “enable JavaScript” notices, placeholder text, and passages repeated within a file or across files. rag-scrubber removes repeated lines at the top and bottom of pages and lines that match its few fixed patterns, and joins split words. Agnos can join split words too, with Re-join words split at a line end, though the RAG / Retrieval Corpus purpose leaves it off.
You want to see what was removed, and why. Review names the rule behind each removed piece, such as “A running header, footer or page number at the edge of a page”. If one file has more than 500 removals, Review counts them all but lists the first 500. Keep this anyway puts a removed piece back; lines such as page numbers come back only by switching their setting off and cleaning again. The Everything export adds rejected.jsonl, with every removed piece, its reason and, for a PDF, its page. rag-scrubber returns only the cleaned text, so checking what it removed means comparing the before and after yourself.
Text in the body of a page must stay. In a PDF or deck with two or more pages, Agnos’s rule for headers and footers looks only at the first two and last two lines of each page or slide, and never removes a line from the body. In a one-page PDF or a one-slide deck, a line that is only a page marker or a “Confidential” banner is removed wherever it is. rag-scrubber removes a line it has flagged wherever it appears, and its patterns remove any line that starts with “Page” and a number or with “Copyright”, in the body too. Its threshold is a share of the pages, so short files need care: in a two-page file at the default of 0.4, one page is enough, and every line of more than three characters among the first and last three of each page is removed.
You want the pages kept with the text. Agnos’s RAG corpus in JSONL gives each piece an id, its source file and its pages, slides, sheet or headings. It also exports cleaned .txt files for Custom GPTs, NotebookLM, Claude Projects or Copilot, and training data with a validation split. rag-scrubber returns one string for the whole document, so keeping track of pages, splitting the text into pieces and writing files are up to your code.
You’d rather pay once than write and look after a script. rag-scrubber is free, and Agnos is not: it is $99 once for one person, or $399 once for up to five people at one company, after a 14-day trial with no card. There is no subscription, and every 2.x update is included.

If you only want to see which lines repeat in one PDF, two free tools need no code. Our Find the repeated lines in a PDF runs in your browser, with nothing uploaded. Elysia Tools’ PDF Header/Footer Noise Remover runs OpenDataLoader on a PDF you upload, once keeping headers and footers and once removing them, and shows the difference.

Both tools are good at their own job. rag-scrubber is a small, free library that takes repeated headers and footers, and page numbers such as “Page 3”, out of page text inside your own Python code; Agnos cleans folders of documents on Windows without code and shows what it took out. They do the same part of the work in different places, so there is little to gain from running both on the same files. If yours is the second job, you can try Agnos free for 14 days, with no card. What it does not do is listed in What Agnos doesn’t do.

Sources

Checked 17 October 2026. rag-scrubber has no website or documentation besides its README. Its PyPI entry links to no repository; the GitHub repository below has the same name, author and README, and holds 0.1.4 release files with the same names and sizes as those on PyPI.

rag-scrubber

rag-scrubber on PyPI, and its metadata: version 0.1.4, the latest, released on 5 January 2026, the same day as versions 0.1.0 to 0.1.3; Python 3.7 or newer; the author; one package for every platform; no licence and no project links; and the README.
MUGESH-KUMAR-M/Rag-Scrubber on GitHub: the README (what it removes, the threshold and its range, the PDF libraries it names, and the licence line), as of the repository’s only commit, on 5 January 2026; no LICENSE file; and the 0.1.4 release files.
cleaner.py, the whole library: the threshold and use_regex_fallback options, the first and last three lines of each page, the fixed patterns, removal wherever a flagged line appears, joined words, the standard library only, and the one string it returns.
setup.py: version 0.1.4, Python 3.7 or newer, no dependencies, and no command line.
The tests: the import from rag_scrubber.cleaner.

Elysia Tools

PDF Header/Footer Noise Remover: listed as a free online tool for PDF files only, which runs OpenDataLoader with headers and footers kept and removed and reports the difference.

Agnos

Supported files, Cleaning settings, Review, Export and Export formats.
Import documents, Your files and privacy, Install and first run, Command line and What Agnos doesn’t do.
Pricing, the guide Remove headers, footers and page numbers from PDF text before RAG, and the free tool Find the repeated lines in a PDF.

rag-scrubber belongs to its author, and Elysia Tools to its owner. This page is ours, not theirs; tell us at hello@agnos.cloud if anything here is out of date.