amal66 is teaching Mike to live inside Microsoft Word
The first, invisible layer of a Word add-in: the backend handshake that lets Mike see the document you have open.
amal66 has started building a Microsoft Word add-in for Mike, and split the work into six reviewable layers. This one is pure backend plumbing, so nothing shows up in Word yet. It teaches Mike's server to accept requests coming from inside Word and to take in the text of the document a user currently has open, so the model can work against what's actually on screen rather than a copy pasted somewhere else.
There's also some early hardening: a cap on how much document text the server will accept at once, and a guard so a missing or malformed document doesn't break the request. The layer was posted for review and then closed without being merged here; amal66 keeps a fully assembled version in their own fork. So this is a stated direction, not a shipped feature.
Spotted something wrong? Or know the PR text has fresher detail than the writeup above?