Front Office redesign · proof

#1338 · after 7c2b4005 vs main 6e485920 · recorded 2026-10-20 · checklist and links

Front Office (new)The redesign, top of the draft stack (7c2b4005)Front Office (before)Today’s main (6e485920), same recorded data Inside an MFL-style pageLoaded the way MFL loads it (fo_embed_loader.js)Trade War Room (Home button)The one War Room route; Front Office links here too ProofBefore/after, reconciliation, tests, what isn’t verified

0 live reads

26-step browser run (Chromium + WebKit): every worker read answered from the recording; 0 reached production.

Every write blocked

POST, PUT, PATCH, DELETE, a Request object, XHR, sendBeacon, form submit, MFL import by GET, and a War Room offer: all stopped before leaving the page.

1442 / 1445

displayed values equal MFL’s recorded exports; the 3 others are exact .x5 halves (below).

222 passed, 0 failed

full safe offline suite on 7c2b4005; 14 of 14 reintroduced bugs caught.

Reconciliation: what the new screens show vs MFL

Read from the preview’s page, then recomputed here from MFL’s own recorded exports with exact arithmetic: points and PPG from playerScores W=ALL (Wks 1–6), the phone’s rank rule (3+ MFL weeks; ties → total points → id), projectedScores W=7, nflByeWeeks, futureDraftPicks, and the league’s waiver check for Status.

Screen	Rows	Values	Result	Differences
Roster · Performance (Real Deal Creel)	37	185	184 match	Brissett, Jacoby PPG: page 19.3, exact 19.4
Market · Free agents (top 200)	200	1200	1198 match	Washington, Darnell PPG: page 8.5, exact 8.6; Bates, Jessie PPG: page 8.0, exact 8.1
Contracts & Cap · Draft picks	60	60	all match

Every rank, bye, projection, status and pick holder matches. The three PPG differences are averages that land exactly on a half (116.1 ÷ 6 = 19.35): the page rounds them down, as the released phone sheet does (same formatter). A shared fix for both is listed as a follow-up.

Earlier reconciliation (issue comment 4)

Rosters: worker vs MFL rosters: 481 / 481 players identical (team, status, salary, years, type).
Cap: Real Deal Creel $286,000 salaries + $9,000 adjustments = $295,000, $5,000 room: identical to MFL’s formula.
War Room cap room used a stale static adjustments file: off by −$33,250 to +$34,000 per team. Fixed in #1340 (live MFL adjustments).
Bye weeks: 0 mismatches; draft picks: all 12 teams identical.

Before / after

Roster

Before (main): contract ledger only; “Pos Rank” is a salary rank.

After: Performance columns (MFL points, PPG, rank, last 4, next game, MFL projection, bye) beside salary; Contract chip keeps the ledger.

Player panel → Stats (same player)

Before: 2026 row from the league database (116, 19.4) mixed into career totals.

After: This season from MFL’s own scoring (116.1, 19.3, #21 QB), as on the phone; earlier seasons separate.

Contracts & Cap

New views

Overview (1440): cap, roster, this week, trade deadline, needs attention, lineup strength vs the league, picks, expiring contracts.

Bye planner: the real 18-slot lineup (SuperFlex, K, P, IDP, D-Flex), Weeks 7–17.

Market: free agents with the league’s waiver status, MFL points, rank, projection, bye.

Trade finder: strength/need matrix, logical partners, fits that help both lineups.

War Room opened from a Trade finder swap: both teams and both players preselected.

Draft picks: who holds every 2027 pick (MFL future picks).

Activity: claims MFL holds (with your MFL sign-in) and 14 days of moves.

Loaded inside an MFL-style page through the real embed loader.

WebKit (Safari engine) render of Trades.

Reintroduced bugs (each must fail its test)

Bug put back	Test	Result	Run
A no write on open	`front_office_no_write_on_open.test.mjs`	caught	1/3 tests passed, 8 assertions, 2 FAILED
B War Room live adjustments	`trade_war_room_live_adjustments.test.mjs`	caught	1/3 tests passed, 5 assertions, 2 FAILED
C player facts = phone	`front_office_player_facts_parity.test.mjs`	caught	5/7 tests passed, 737 assertions, 2 FAILED
D shell + deep links	`front_office_shell_nav.test.mjs`	caught	Node.js v22.22.2
D rank waits for players list (old hub)	`front_office_rank_universe.test.mjs`	caught	1/4 tests passed, 7 assertions, 3 FAILED
D rank cache keyed on groupOf (old cache)	`front_office_rank_universe.test.mjs`	caught	3/4 tests passed, 18 assertions, 1 FAILED
D bye planner: IR can start	`bye_planner_model.test.mjs`	caught	4/5 tests passed, 19 assertions, 1 FAILED (after adding the IR-designation case; missed before)
E blank projection read as 0	`front_office_roster_performance.test.mjs`	caught	4/5 tests passed, 51 assertions, 1 FAILED
F claim-now shown as add-now	`front_office_market.test.mjs`	caught	3/5 tests passed, 25 assertions, 2 FAILED
G kickers/punters used in fits	`trade_fit_model.test.mjs`	caught	7/8 tests passed, 75 assertions, 1 FAILED
H swap link drops the players	`front_office_trade_finder.test.mjs`	caught	3/4 tests passed, 12 assertions, 1 FAILED
H War Room ignores give/get (main)	`trade_war_room_fit_link.test.mjs`	caught	0/3 tests passed, 1 assertions, 3 FAILED
I IR moves read backwards	`front_office_picks_activity.test.mjs`	caught	2/3 tests passed, 10 assertions, 1 FAILED
I fixed 1-6 pick columns	`front_office_picks_activity.test.mjs`	caught	2/3 tests passed, 13 assertions, 1 FAILED

Not verified here

Signed-in MFL behavior: claims MFL holds, owner-only actions with a real session, and the commissioner view.
Real Safari on a Mac (WebKit engine only) and the live MFL page’s own header and scripts around Front Office.
Live freshness: the preview replays one recording; the live site reads MFL’s exports (60-second cache).
Worker change in #1341 (player-status id map cache) runs only in the worker; covered by its offline test, not this preview.