Read us freely. Cite us properly. Licence us to absorb us.

The reflex when protecting a data asset is to shut the crawler out. That would be the wrong move here, and we want to be explicit about why we have not done it.

This platform exists to be cited. A corpus nobody can read is a corpus nobody cites, and every named answer engine keeps unrestricted access to every document, the full-text corpus, and every JSON endpoint. We restrict the use, not the access.

robots.txt cannot express this. It answers one question — may this crawler fetch this URL — and has no vocabulary for purpose. The distinction that matters to us is "read and cite freely, but do not absorb into a commercial product," which is why the terms live here and in ai.txt rather than in a crawler block.

Attribution means a hyperlinked reference to the canonical URL declared in each page's <link rel="canonical"> — not the domain root. Where a figure is reproduced, reproduce the attribution that accompanies it; the underlying estimates belong to the institutions named on the page, not to us.

Revocability. The free grant is perpetual for uses already made in reliance on it. We will not retroactively invoice a citation, and nothing here should be read as reserving the right to.

We want to be precise about this, because overclaiming would be worse than not claiming at all.

Not ours: the raw market inputs. Index levels, volatility readings and factor returns are public facts, and facts are not owned by anyone. The published regulations, standards and specifications described on this platform belong to the bodies that issued them, and we link to every primary source. The vendors and tools in our registers are their own.

Ours: the taxonomy. Which concepts exist and how they are bounded. Which entities are included and on what criteria. The typed edges between them — the assertion that this concept depends on that one, that this standard addresses this governance question and not that one. The mapping of 230 control objectives onto eight questions. The scores, the caveats, and the editorial judgements throughout.

Selection and arrangement are the asset. A competitor is entirely free to look at the same public market data and the same public standards and build their own map. They are not free to take ours.

Sustained bulk transfer of the full graph is rate limited at the edge. This is a throttle, not a gate: it exists to distinguish a reader from a mirror, and no legitimate crawler, researcher or developer will encounter it. Ordinary crawling — including a complete crawl of all 100-plus documents — is unaffected.

If you hit a limit and have a good reason to need the whole thing at speed, ask us. We will almost certainly say yes and hand you a faster route.

If your use falls in the licence-required column, the conversation is short and we are not difficult to deal with. Academic and non-commercial work is free — if you are unsure which side you are on, that is itself a reason to write.

The terms above are also published machine-readably: inline in an api_metadata block at the top of every JSON endpoint, in /ai.txt and /.well-known/ai.txt, and in the header of llms-full.txt. There is no version of accessing this platform that leaves you unaware of the terms.