fix(docs): 49 stale numbers three grep passes had missed - #22
Merged
Conversation
#18, #19 and #21 each swept this tree by hand for whichever phrasings came to mind, and each found a different subset. What none of them found: 64 models / 64 LLMs 29 places — the catalog serves 66 84 AI models 9 places — 86 across every modality a free tier of 10 4 places — 8 are published; the rest are withheld Save 78% / up to 94% 7 places — two more savings figures, five in total The claim SHAPES varied, not the numbers, which is exactly what someone scanning for "64" will not notice. Found by sweeping for the shape — a number in front of the claim vocabulary — instead of for remembered phrasings. blockrun now runs that sweep as a test over this tree on every PR, so the next one fails there rather than shipping. Numbers that only look like claims are exempted by name with a reason, so the allowlist cannot quietly grow into a no-op. README's last unmarked savings claim gets a marker. The docs tree cannot take them: blockrun renders it and escapes the comment into view. docs/sdks/xrpl.md is deliberately untouched. It documents the XRPL gateway, decommissioned 2026-06-06 — restating its numbers would be polishing a page that should probably not be published at all. Flagged, not fixed.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
#18, #19 and #21 each swept this tree by hand, each for whichever phrasings came to mind, and each found a different subset. Here is what none of them found:
64 models/64 LLMs84 AI models10 modelsSave 78%/up to 94%49 in total, all live on blockrun.ai/docs — this tree is symlinked into blockrun and served directly.
up to 94%is the fifth distinct savings figure this project has published.Why three passes missed them
The claim shapes varied, not the numbers. Someone scanning for
64does not notice84 AI models, and nobody thinks to grep for "a free tier of 10".So this pass swept for the shape — a number in front of the claim vocabulary — rather than for phrasings anyone remembered to write down. That found 10× what hand-grepping did.
It is now a test, not a habit
blockrun runs that sweep over this tree on every PR: any
N models|LLMs|tools|chainsor savings percentage that is not a published value fails, with the file and line. Verified non-vacuous — restoring one64 modelsfails withdocs/resources/faq.md:105.Numbers that only look like claims are exempted by name with a reason, so the allowlist cannot quietly grow until the sweep tests nothing:
docs/sdks/xrpl.md— decommissioned gatewayREADME
Its last unmarked savings claim gets a marker. The
docs/tree still cannot take them — blockrun renders it and escapes the comment into view (#21).Deliberately not fixed
docs/sdks/xrpl.mddocuments the XRPL gateway, decommissioned 2026-06-06, and is still published. Restating its numbers would be polishing a page that probably should not exist. That is a question about the page, not the numbers — flagging rather than deciding.