Update to Seal Copilot

{
  "description": "The single proactive report the customer sees first thing every Monday: weekly performance verdict + top opportunity + watchdog status, on one screen, email-shareable. Trigger on: \"monday briefing\", \"morning report\", \"weekly briefing\", \"informe del lunes\", \"start my week\", \"what should I do this week\", \"lunes\", \"executive briefing\", or when run from the scheduled Cowork job.\n",
  "included_files": [
    {
      "relative_path": "examples/output.md",
      "size_in_bytes": 1472
    }
  ],
  "name": "monday-briefing",
  "skill_md_contents": "---\nname: monday-briefing\ndescription: >\n  The single proactive report the customer sees first thing every Monday:\n  weekly performance verdict + top opportunity + watchdog status, on one\n  screen, email-shareable. Trigger on: \"monday briefing\", \"morning report\",\n  \"weekly briefing\", \"informe del lunes\", \"start my week\", \"what should I\n  do this week\", \"lunes\", \"executive briefing\", or when run from the\n  scheduled Cowork job.\ndisable-model-invocation: true\nshort-description: 'The one-page Monday briefing: verdict, week vs last, what worked, one opportunity, watchdog. Use for \"Monday briefing\", \"briefing\", \"resumen del lunes\", or a scheduled run.'\n---\n\n# Monday Briefing\n\nBefore writing your answer, read `examples/output.md` in this skill directory\nand match its density, structure and tone. It is the reference for what a good\nrun of this skill looks like.\n\nThe Monday-morning one-pager. Combines the highlights of three skills\ninto a 6-block report the user can forward to their team. Budget: ≤15\ncalls. Designed for **scheduled execution** — `/schedule` in Claude Code, or\nthe equivalent scheduled task in Cowork (\"every Monday at 8 am in [site\ntimezone]\").\n\n## Composition\n\nThis skill **calls down** to three other skills' procedures but runs them\nin compact mode — do not produce three full reports, produce one merged\nreport. Each section caps at the line counts below.\n\n1. Weekly health check — slim (KPIs + verdict only, no findings).\n2. Opportunity scan — top 1 only (the highest-€ pattern that fires).\n3. Cart-watchdog — current status only (no historical baseline detail).\n\nIf any sub-step errors, show \"—\" for that section, do not abort the rest.\n\n## Procedure\n\n### Block 0 — Follow-up (0–2 calls)\nRead `<state-dir>/<site_id>/recommendations.jsonl`. For entries with\n`status: open` and `verify_on` today or earlier, re-measure and mark them\nverified or failed (see `skills/seal-copilot/references/state-schema.md`).\nAt most two re-measurements per briefing — the rest wait a week. If nothing\nis due, this block produces no output at all.\n\n### Block A — Snapshot (3 calls)\n1. `get_overview(period=7d, compare=previous)` (for hotels also\n   `compare=yoy`).\n2. `get_top_channels(period=7d)` — ranked, no `compare` available.\n3. `get_campaigns(period=7d, sort_by=revenue, limit=5)` — use the full tool\n   when you need sorting; `get_top_campaigns` is ranked by entrances only.\n\n### Block B — Opportunity radar (3–5 calls)\nPick the **single highest-impact pattern** that fires from this short list\n(do not run the full pattern library):\n- Hidden star campaign (top CR + AOV, low volume) — `get_campaigns(\n  sort_by=revenue, limit=20, compare=previous)`.\n- Leaky campaign (high entrances, low CR) — same call, opposite end.\n- Mobile gap (mobile CR <50% desktop) — `get_device_types(period=7d)`.\n- Untapped country (high CR, no campaign) —\n  `get_countries(period=30d, sort_by=conversions, limit=10)`. Country is\n  timezone-derived: report it as a question, not a spend recommendation.\n\nPick one only — the one with the largest € impact. Do not list the others.\n\n### Block C — Watchdog status (2 calls)\n- `get_microconversions(conversion_type=<atc-equivalent>, period=today)` —\n  today's volume so far, compared against the stored watchdog baseline if\n  `calibrate-watchdog` has run. Without a baseline, compare to\n  `get_microconversions(conversion_type=<atc>, period=yesterday)` and say the\n  comparison is coarse.\n- `get_bot_stats(days=7)` — bot share trend. Empty means agent analytics is\n  off, not 0%.\n\nStatus line: `🟢 normal` / `⚠️ watch — <reason>` / `🔴 act now — <reason>`.\n\n### Block D — Validation (0 calls)\nReuse the `get_bot_stats(days=7)` result from Block C. If it was empty or\nreturned 403, mark every mover in Block A \"unvalidated for bots\".\n\n## Output format (the one-pager)\n\n```\n🦭 Seal Copilot — Monday Briefing · <site> · <week dates>\n\n🎯 VERDICT: <✅ on track | ⚠️ watch | 🔴 act now>\n<one sentence summarizing the week>\n\n📊 THIS WEEK vs LAST\n| Entrances | CR | Conversions | Revenue | AOV |\n|  X (±%)   | X% | N (±%)      | €X (±%) | €X  |\n(for hotels: same row vs yoy underneath)\n\n🏆 WHAT WORKED\n<the single best channel or campaign this week, with numbers>\n\n🩹 WHAT NEEDS ATTENTION\n<the single biggest negative mover, with numbers and likely cause>\n\n💰 TOP OPPORTUNITY THIS WEEK\n<pattern name>: <evidence> → <action> → <est. € impact> → <verify in 2-4 wk>\n\n✅ FOLLOW-UP\n<only if something was due: one line per recommendation checked, with the\nnumber that moved and verified/failed. Omit the whole block if nothing was due.>\n\n⛔ NOT CHECKED\n<only if a step's call was refused or skipped: one line naming it, e.g.\n\"channel split and bot validation — API refused get_channels / get_bot_stats\nfor this site; movers above are unvalidated for bots\". Omit if all ran.>\n\n🚨 WATCHDOG\nAdd-to-cart: <🟢/⚠️/🔴 + one-line context>\nTracking decay (microconversions): <🟢/⚠️/🔴>\nBot share: X% (last week Y%)\n\n➡️ NEXT\nSuggested follow-up: \"<one concrete next prompt the user can paste>\"\n```\n\nKeep the whole output under ~30 lines so it copy-pastes into email/Slack\ncleanly. No code blocks except the verdict box. No filler.\n\nAppend the opportunity you reported to `recommendations.jsonl` and log the\nrun in `runs.jsonl` with exactly `ts`, `skill`, `calls`, `budget`, `verdict`,\n`scheduled`, `notes` — `budget` is `15` for this skill, `calls` is counted.\n\n## Scheduling guidance\n\nOn first successful run, offer:\n\n> \"Want this every Monday at 8 am? Reply 'schedule monday-briefing' and I\n> will set it up.\"\n\nWhen the scheduler fires this skill, the output is the entire response —\nno preamble, no \"Hi! Here is your briefing\", just the one-pager above.\n\n## What you do NOT do\n\n- Do not include >1 opportunity. Monday is for focus.\n- Do not run the full opportunity-scan, full health-check, or full\n  watchdog procedures here — call them by name as follow-ups if the user\n  wants depth.\n- Do not include statistical caveats inside the one-pager; if a number is\n  low-confidence, suffix it with \"(low sample)\".\n- Do not personalize the verdict beyond the data — no \"great job\" / \"bad\n  week\" framing.\n"
}