Across the sgit.ai family, most writing starts as a markdown brief. Some of those briefs get turned into a single visual artefact — usually by pasting the brief into an image model by hand, not by any site's own build pipeline. Once that image exists it tends to live nowhere durable: a chat thread, a downloads folder, a Slack message. infographics.sgit.ai is where it lives instead — stored, dated, and kept traceable back to the brief that argued for it and the document that rendered it in prose.
Three steps, none of them automated end to end — that is a stated limit, not an oversight. See comms for what would change that.
A markdown file in a sibling site's briefs/ folder, a vault document, a strategy note. It is already the source of truth before any image is made.
Today, almost always a person pasting the brief into an image model by hand. A documented pipeline exists for this at scale — see the network page — but most entries so far were not made with it.
The image, plus a metadata.json naming the source brief, the source site, what rendered it, and its licence. Nothing is catalogued without that trail.
What this site is not. It is not an infographic generator, and it does not decide what a brief means — the brief and its rendered document (where one exists, usually on the source site's own documents/ section) stay the source of truth. This site is a library card catalogue: it stores the picture, and points back at the argument.
infographics.sgit.ai stores rendered visual artefacts and the metadata linking each one back to its source brief and, where one exists, its rendered prose document on the source site. It does not itself generate infographics, and a picture appearing here is not evidence that the source site's own build pipeline produced it — check each entry's generator field. library/data/catalogue.json is the full machine-readable index. llms.txt is the whole agent surface.