About BrainyWorks

BrainyWorks is a small set of document tools that do one thing each, do it quickly, and forget your file the moment they are done.

Why it exists

Free PDF tools are not scarce. Trustworthy ones are. Most upload your document to a server and keep it for anywhere between an hour and forever, which is fine for a holiday itinerary and not fine for a contract, a payslip, or a discharge summary.

Everything here processes files in memory and drops them immediately. No account, no email gate, no watermark, no daily limit, and no filename written to any log.

The scanner

The Document Injection Scanner is the reason this site exists at all. Document AI pipelines now read incoming files with a language model, and in most of them there is no boundary between "text extracted from the page" and "instructions the model should follow." A document can carry text a human reviewer will never see — white on white, one point tall, hidden under an image, tucked into a metadata field — that the model reads as an instruction.

The scanner looks for exactly that, and reports hidden content separately from hidden content that reads like an instruction, because those are very different findings. It is open about its limits: heuristics miss novel phrasings, and a clean result means nothing was found, not that nothing is there.

Who builds it

Built and run by Aditya Mangal, an AI engineer in Jaipur working on document processing — vision language models, retrieval, and extraction pipelines for regulated industries. The tools here come out of that work: the problems are real ones met in production, reduced to something anyone can run in a browser.

Suggestions and bug reports are welcome at hello@brainyworks.in. More tools are added as they prove useful.

How it is built

Python and PyMuPDF for the document work, FastAPI for the server, and plain HTML and CSS for the pages. No third-party scripts, no CDN, no analytics SDK — every asset is served from the same box, which is what makes the privacy claims on this site checkable rather than aspirational.

See the privacy page for exactly what is and is not collected.