Side tools: RAG mbox viewer

Nov 7th, 2025

I am working recently a lot on RAG pipelines, and I needed to inspect a corpus stored in an mbox-like format, so I wrote a small tool for it.

The tool is very simple and is written as a few small ksh93 scripts.

It happens that I had always wanted a lightweight way to read archival messages from Usenet, but being a historical curiosity it was never a priority. With just a couple of minor changes, the same tool could also read actual Usenet mbox files.

Usenet was the main discussion and Q&A network before the Internet exploded in the mid-90s. It ran on top of UUCP, a store-and-forward protocol from the early 1980s that let UNIX hosts exchange files and messages over dial-up links once or twice a day.

You can find early messages and historical announcements there. If you read them, you’ll immediately recognise a much more civilised Internet, most of the users came from universities or technology-related companies.

Side tools are often underestimated. Many see them as liabilities compared to modern cloud alternatives, but they’re mature, dependency-free, and require no maintenance. Standard tools are also rarely fit for purpose, so you end up having to build something much bigger.

I share here the Usenet side project of the tool rather than the actual one, because it is far more interesting, especially for those who might have lived through that period in time.

2025-11-07