Word addin spike - #90
Merged
Merged
Conversation
It works, so it should be findable by somebody who is not reading pull requests. Both sections lead with the argument rather than the feature. A task pane is a browser, so the engine runs on the machine and the document is never uploaded — in the application where the document already lives. Every other add-in of this kind posts your text to an API because their analysis is a server; ours is not, so there is nothing to post it to. And the manifest asks for ReadDocument rather than ReadWriteDocument, which means Word enforces "reads only" instead of anyone being asked to trust a sentence on a website. The README section shows what it does on a real document, and the example is the 357-word one from the first working sideload: a score, no verdict, the reason printed, and six no-break spaces reported underneath. Both halves are the point. It refused to answer at that length because the corpus has no text shorter than 649 words, and it still reported the characters, because the scan is a fact about the file and carries no threshold. Also says plainly that PowerPoint is a different product rather than a second Host entry: a deck rarely reaches 649 words, so the same add-in there would mostly withhold the verdict — correctly, and to little use. The site section goes on /download, after the desktop app rather than beside it: that page is already "run it on your own machine", and this is the other way to do that. New `file-text` glyph in Icon.razor, since the house rule is a drawn line-icon and never an emoji. 432 tests. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015PEbbiYSNPw7jE3LrPNhyF
Pedro's screenshot, because the address bar is in it: word.cloud.microsoft, a real document, and the pane answering beside it. That settles in one image what three paragraphs were arguing. His OneDrive docId and driveId are redacted and the browser's tab strip is cropped. The domain stays, because the domain is the point. The prose around it went through our own linter, which is the house rule for anything published, and it charged the section with em-dash overuse: 10 in 550 words, 1.8 per 100, Medium. Fair, and worth acting on the day after shipping the fix that stopped that rule counting markup — these were real dashes. Four changed because they read better, not to move a number: one introducing a list became a colon, one long apposition became a second sentence, a doubled parenthetical became parentheses, and one sitting in front of "and" became a comma. The five left are doing work — a heading's subtitle, a label, a figure's apposition, one genuine parenthetical pair. Section down to 9/100 with the em-dash finding gone; it was never the target, and if a rewrite had read worse it would have stayed. 432 tests. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015PEbbiYSNPw7jE3LrPNhyF
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.
What this changes
Why
Checklist
dotnet testpasses, with no new build warningsIf you touched the rule packs (
rules.en.json/rules.es.json):suggestion— what the writer should do insteadtests/SignsOfAI.Core.Testscovering itCleanHumaninScoringTestsstill scores exactly 0 (the new rule doesn't fire on ordinary prose).*; this runs on every keystroke in the browserIf you touched the articles in
Docs/Blog: