export declare const VALIDATE_CHECK = "# /validate-check \u2014 Daily 60-Second Health Check\n\n## What This Does\n\nRapid daily diagnostic during active validation. User pastes raw numbers, gets an instant traffic-light verdict and one specific action. Not a full analytics review \u2014 just the minimum viable feedback loop.\n\n## Trigger Context\nDuring any active validation window. User runs this daily (or every other day minimum).\n\n## Execution Protocol\n\n### Step 1: Accept Data Input\n\nAccept data in ANY format. The user might:\n- Paste the daily template from `/validate-prep`\n- Type numbers conversationally (\"spent $47 today, got 89 clicks, 3 signups\")\n- Describe a screenshot (\"Meta shows CTR 1.2%, CPC $0.53, 2 conversions\")\n- Paste a CSV or table from a dashboard\n\nParse whatever comes in. Don't ask for reformatting. Extract what's available, flag what's missing only if it's critical to the diagnosis.\n\n### Step 2: Load Context\n\nRead from pipeline-state.json:\n- Current validation tier\n- Decision thresholds (from /validate-prep)\n- Previous daily check data (if any)\n- Unit economics guardrails\n- Days remaining in validation window\n\n### Step 3: Instant Diagnosis\n\nCompare today's data against thresholds. Output ONE of three verdicts:\n\n**\uD83D\uDFE2 ON TRACK**\n```\nDay [X] of [Y] | Spend: $[cumulative] of $[budget]\nKey metric: [the one that matters most at this tier] = [value] (target: [threshold])\nStatus: On track. No changes needed.\nAction: Keep running. Check again tomorrow.\n```\n\n**\uD83D\uDFE1 WATCH**\n```\nDay [X] of [Y] | Spend: $[cumulative] of $[budget]\nConcern: [specific metric] at [value] (target: [threshold])\nContext: [one sentence \u2014 is this normal early noise or a real signal?]\nAction: [one specific thing to check or adjust \u2014 NOT a full overhaul]\nCheck again: [tomorrow / in 48 hours]\n```\n\n**\uD83D\uDD34 KILL SIGNAL**\n```\nDay [X] of [Y] | Spend: $[cumulative] of $[budget]\nProblem: [specific metric] at [value] \u2014 hit kill threshold of [threshold]\nDiagnosis: [one sentence root cause \u2014 creative, targeting, page, or offer]\nAction: PAUSE [specific campaign/ad]. [One specific next step.]\nRun /validate-decide now? Or give it [X] more hours?\n```\n\n### Step 4: Trend Detection\n\nIf this is day 3+, compare against previous daily checks:\n- Improving trend \u2192 Note it. \"CPA dropping steadily: $12 \u2192 $9 \u2192 $7. Trending toward target.\"\n- Degrading trend \u2192 Flag it. \"CTR declining 3 days straight. Creative fatigue starting early.\"\n- Volatile / no pattern \u2192 Normal for small budgets. Note it but don't overreact.\n\n### Step 5: Update State\n\nAppend today's data to pipeline-state.json validation log:\n```json\n{\n  \"date\": \"[today]\",\n  \"day_number\": [X],\n  \"data\": { [whatever was provided] },\n  \"verdict\": \"[green/yellow/red]\",\n  \"action_taken\": \"[what was recommended]\",\n  \"cumulative_spend\": [running total],\n  \"cumulative_conversions\": [running total]\n}\n```\n\n### Step 6: Countdown\n\n```\nValidation window: [X] days remaining.\nNext check: [tomorrow's date]\nDecision date: [date from /validate-prep]\n```\n\n## Rules\n- NEVER recommend a full campaign overhaul from a single day's data at small budgets.\n- At $50/day, daily fluctuations are noise. Trends over 3+ days are signal.\n- Don't overwhelm with metrics. One verdict, one number, one action.\n- If data is insufficient to diagnose (e.g., only 12 impressions), say so: \"Not enough data yet. This is normal on day 1. Check again tomorrow.\"\n- Always end with the countdown to decision date.\n\n## Autonomy\nTier 1 \u2014 Full Auto. No decisions required. Pure information delivery.\n\n## Output\nTraffic light verdict + one action. Presented inline, no file saved (state updated in pipeline-state.json).\n\n## Next Step\nTomorrow \u2192 `/validate-check` again\nDecision date reached \u2192 `/validate-decide`\nKill signal confirmed \u2192 `/validate-decide` immediately";
//# sourceMappingURL=validate-check.d.ts.map