Verification and Constraints

These are exact launch defaults. A deviation must name the evidence and changed field in the board.

Preserve useful existing exclusions. Add only missing terms and check for conflicts.

The current custom goal and its action IDs were read from the account. Tests have not passed yet.

Allocation: drilling 1,800 + concrete 1,700 + paving 1,300 + turf 450 + irrigation 450 + maintenance 800 = 6,500 Ft/day. Priorities receive 4,800 Ft/day (73.85%). Each campaign has its separate existing non-shared budget. Maintenance’s 800 Ft is shared by two groups. Concrete/paving are changed from the current 1,300/1,700 allocation to follow the stated priority order.

For a full month with these unchanged budgets, Google’s usual 30.4× monthly charging bounds sum to 197,600 Ft. The 2,400 Ft difference is headroom, not extra spend to allocate. This is not a native hard account cap and does not cover earlier higher budgets automatically.

Before activation or ANY budget edit, calculate an account-wide upper bound: cost through a finalized prior day + a conservative charge bound for every unfinalized day + future permitted charges. On a change day use 2× each budget’s highest value that day. Count shared budgets once. For remaining changed-month days, the conservative 2×daily upper bound is acceptable. If earlier history cannot establish a bound, affected spend remains paused.

September 25–30 example, only with a fresh ledger proving zero earlier cost/exposure: because concrete rises 1,300→1,700 and paving falls 1,700→1,300 on the same day, per-campaign highest budgets sum to 6,900. September 25 bound = 2×6,900 = 13,800, plus September 26–30 = 5×2×6,500 = 65,000, total 78,800 Ft. The 78,000 Ft figure applies only to six unchanged 6,500 Ft days, not this swap day. Recompute for the actual launch date/history, do not catch up unused spend. Next full month remains 6,500/day.

Boundary fixtures: (A) six unchanged full days, prior finalized cost 130,000:130,000+78,000=208,000, reject 6,500/day; floor70,000/12=5,833/day before round-down allocations. (B) swap-day high-water 6,900/day already established:130,000+13,800+5×13,000=208,800, reject. Reserve the full 13,800 for that day, leaving 56,200 for five future days, hence total daily allocation ≤5,620 then, or pause. A same-day reduction does not erase its earlier highest budgets. The live worksheet must compute each campaign/day maximum from history before choosing scaled budgets.

Executor owns fresh preflight, temporary pause of these six during edits, before/after ledger and 24h/48h checks. Record exact checkpoint times and next-month baseline now. A scheduled month-boundary recheck remains pending until performed, not claimed complete. Do not build a spend controller. No budget increase or reactivation outside this six-campaign plan. Matrix amounts are maximum baseline allocations, permitted proportional reductions must be recorded consistently in settings-diff and budget-ledger.

Provider rules: Budgets overview · Budget changes · Maximize Clicks caps · Campaign conversion goals

Every GUI test needs a screenshot of the actual changed interface after the stated test. An independent gpt-6-luna high reviewer receives the screenshot without the expected result and describes only what is visible. Attach screenshot, blind description and parent comparison to that criterion. Missing or mismatched proof leaves it unchecked. Provider exports stay private; publish only redacted summaries/screenshots. Setup/readiness and subsequent serving are separate stage results. The proposed whole-goal acceptance retains serving proof, not a promise that auctions produce impressions within 48 hours. At a checkpoint, zero remains honestly open with diagnosis and next action.

Open, to verify: The cause of LIMITED/UNKNOWN and zero delivery still needs provider diagnostics. Non-maintenance geographic coverage is retained from the account, not newly agreed in the meeting. The 600/400 Ft CPC limits are reviewable starting choices, not performance estimates. Only the maintenance town targeting point is verified, not a home address. Live form/attribution proof is still incomplete. A failed gate stays unchecked.

Assumption: existing campaigns and pages can be corrected without a broader rebuild. The executor checks this and reports any specific campaign dependency.

Only criteria with complete receipts are checked. Remaining criteria retain a precise pending state or blocker. A configured campaign is not a serving campaign.

LAUNCH AUTHORIZED in Matt’s latest message: create one NEW visible gpt-6-sol/high Ads execution thread for this reviewed Google-only contract, with the maintenance destination correction. A separate parent-owned Sol Medium worker implements and verifies the public UI changes; Ads executor does not edit those files and waits for verified live /kertfenntartas before sending traffic there. Review comments and inline rewrites were checked on 25 September (zero comments at launch preparation) and any later feedback must be reconciled. No client sends, changed commercial commitments, or more than USD15 shared vendor spend. New review/deployed surfaces use bare publicly reachable URLs with no added security gate. Keep credentials and sensitive source data private.

Executor messages will appear here after Launch.

Private executor inputs: executor-source-map.md and the reviewed contract. On completion: deliverables.md, main-prompts.md, session.txt and a clean gist. Update this board and its proof counters on the same URL.