The shell split the path itself and chose a page with a chain of ifs, and every page began reading on render. The router names each page as a route with the read it needs as its loader, warming the query the page then subscribes to under the same key: the answer is in hand before the page renders, a pull request's diff, commits and statements are read together rather than one after the other, and a link starts its page's read when the pointer settles on it. The addressing the pages own stays theirs. A pull request's address holds a DID and a branch with slashes in it, and a ref may hold slashes before a path, so those routes take the rest of the path whole and parse it with the same functions as before. The navigation module is an adapter over the router in the shape the pages take, so a page calls what it called. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RomHvaB4nyAckbRD7W36Ue