@wonsukchoi/crondex 0.7.0 → 0.9.0

package/jobs/gaming/subscriber-churn-watch.yaml

@@ -0,0 +1,39 @@
+id: subscriber-churn-watch
+version: 1
+name: Subscriber/Follower Churn Watch
+description: >
+  Checks your channel's subscriber or follower count trend and flags a
+  meaningful drop so you can investigate before it becomes a pattern. Use
+  this if you've ever noticed a slow bleed of subscribers only after
+  checking your dashboard months later.
+category: gaming
+tags: [streaming, subscribers, churn, analytics]
+schedule: "0 9 * * 1"
+timezone: "America/Los_Angeles"
+runner: agent-prompt
+prompt: |
+  Check subscriber/follower counts for {{channel_name}} on
+  {{platform_name}} for the last {{lookback_weeks}} weeks (via any
+  analytics dashboard or API available).
+  1. Report week-over-week change in subscriber/follower count.
+  2. Flag if this week's net change is negative or more than
+     {{drop_threshold_pct}}% below the recent average.
+  3. If there's a drop, note anything that coincided with it (missed
+     stream days, controversial clip, platform-wide event) if that
+     information is available.
+  4. If growth is healthy, just report the trend in one line.
+variables:
+  channel_name:
+    default: "your channel"
+    description: Channel/handle to check.
+  platform_name:
+    default: "Twitch"
+    description: Platform to check (e.g. "Twitch", "YouTube", "Kick").
+  lookback_weeks:
+    default: 8
+    description: Weeks of history to establish a baseline trend.
+  drop_threshold_pct:
+    default: 15
+    description: Percentage drop below the recent average that counts as worth flagging.
+compatible_agents: [claude, codex, hermes, openclaw, generic]
+notes: Depends on what the platform's API/dashboard exposes — some platforms require OAuth to pull subscriber history.

package/jobs/healthcare/appointment-no-show-follow-up.yaml

@@ -0,0 +1,32 @@
+id: appointment-no-show-follow-up
+version: 1
+name: Appointment No-Show Follow-Up
+description: >
+  Reviews yesterday's no-show appointments and drafts a follow-up message
+  offering to rebook. Use this if you've ever let a missed appointment go
+  quiet and lost the patient/client entirely instead of just rebooking them.
+category: healthcare
+tags: [appointments, no-show, patients, scheduling]
+schedule: "0 9 * * *"
+timezone: "America/New_York"
+runner: agent-prompt
+prompt: |
+  Pull yesterday's appointments from {{scheduling_system}} and find any marked
+  no-show or cancelled-late.
+  1. For each, note the patient/client name, appointment type, and provider.
+  2. Draft a short, warm follow-up message (SMS or email length) offering to
+     rebook, without shaming them for missing it.
+  3. Flag anyone with 2+ no-shows in the last {{lookback_days}} days
+     separately — practices often want a different (firmer) policy for
+     repeat no-shows.
+  4. Output as a list ready for staff to review and send — do not send
+     anything automatically.
+variables:
+  scheduling_system:
+    default: "your scheduling/EHR system"
+    description: Name of the scheduling or EHR system to check (e.g. "Epic", "Calendly", "SimplePractice").
+  lookback_days:
+    default: 90
+    description: Window to check for repeat no-show patterns.
+compatible_agents: [claude, codex, hermes, openclaw, generic]
+notes: Draft-only — never send messages automatically. Confirm HIPAA/compliance rules for your practice before wiring this to an agent with system access.

package/jobs/healthcare/license-credential-expiry-check.yaml

@@ -0,0 +1,35 @@
+id: license-credential-expiry-check
+version: 1
+name: Staff License/Credential Expiry Check
+description: >
+  Checks a tracked list of staff licenses, certifications, or credentials
+  (e.g. nursing license, CPR cert, DEA registration) and warns before any
+  expire. Use this if you've ever had a provider accidentally see patients
+  on a lapsed license or certification.
+category: healthcare
+tags: [compliance, licensing, credentials, staff]
+schedule: "0 8 * * 1"
+timezone: "UTC"
+runner: shell
+command: >
+  awk -F',' -v days="{{warn_days_before}}" '
+    NR==1 {next}
+    {
+      cmd = "date -d \"" $3 "\" +%s 2>/dev/null || date -j -f %Y-%m-%d \"" $3 "\" +%s 2>/dev/null";
+      cmd | getline expiry_epoch; close(cmd);
+      "date +%s" | getline now_epoch; close("date +%s");
+      diff_days = (expiry_epoch - now_epoch) / 86400;
+      if (diff_days <= days) {
+        printf "EXPIRING: %s — %s expires %s (%d days)\n", $1, $2, $3, diff_days;
+      }
+    }
+  ' "{{credential_csv_path}}"
+variables:
+  credential_csv_path:
+    default: "./staff-credentials.csv"
+    description: "Path to a CSV of staff credentials with columns: staff_name,credential_type,expiry_date (YYYY-MM-DD)."
+  warn_days_before:
+    default: 45
+    description: How many days before expiry to start warning.
+compatible_agents: [generic, claude, codex]
+notes: Assumes a maintained CSV export from your credentialing/HR system — point it at whatever that system can export on a schedule.

package/jobs/healthcare/patient-recall-reminder.yaml

@@ -0,0 +1,34 @@
+id: patient-recall-reminder
+version: 1
+name: Patient Recall Reminder
+description: >
+  Finds patients due (or overdue) for a routine checkup, cleaning, or
+  screening and lists them for outreach. Use this if recall/recare lists
+  pile up and patients quietly fall off the schedule for a year or more.
+category: healthcare
+tags: [patients, recall, checkup, outreach]
+schedule: "0 8 1 * *"
+timezone: "America/New_York"
+runner: agent-prompt
+prompt: |
+  Check {{scheduling_system}} for patients whose last visit for
+  "{{recall_type}}" was more than {{recall_interval_months}} months ago and
+  who have no future appointment booked.
+  1. List each patient, their last visit date, and how overdue they are.
+  2. Sort by most overdue first.
+  3. Draft a short recall message per patient (or a template if the list is
+     long) inviting them to book.
+  4. Call out anyone more than double the recall interval overdue — these
+     are the ones most at risk of having left the practice entirely.
+variables:
+  scheduling_system:
+    default: "your scheduling/EHR system"
+    description: Name of the scheduling or EHR system to check.
+  recall_type:
+    default: "annual checkup"
+    description: What kind of visit this recall tracks (e.g. "dental cleaning", "annual physical", "eye exam").
+  recall_interval_months:
+    default: 12
+    description: How many months between visits before someone counts as due.
+compatible_agents: [claude, codex, hermes, openclaw, generic]
+notes: Draft-only — outreach messages should be reviewed by staff before sending, and must follow your practice's patient communication/compliance policy.

package/jobs/hiring/offer-expiry-reminder.yaml

@@ -0,0 +1,31 @@
+id: offer-expiry-reminder
+version: 1
+name: Offer Expiry Reminder
+description: >
+  Flags extended offers that are close to their response deadline with no
+  answer yet. Use this if an offer has ever quietly expired because
+  nobody followed up before the deadline passed.
+category: hiring
+tags: [hiring, offers, recruiting]
+schedule: "0 9 * * 1-5"
+timezone: "UTC"
+runner: shell
+command: >
+  today=$(date +%s);
+  awk -F, -v today="$today" -v warn={{warn_days}} '
+    NR>1 && $3 == "pending" {
+      cmd="date -d \"" $2 "\" +%s 2>/dev/null || date -jf \"%Y-%m-%d\" \"" $2 "\" +%s";
+      cmd | getline exp_ts; close(cmd);
+      days_left = int((exp_ts - today) / 86400);
+      if (days_left <= warn && days_left >= 0) print "WARNING: offer to " $1 " expires in " days_left " days (" $2 "), no response yet";
+    }
+  ' {{offers_csv}}
+variables:
+  offers_csv:
+    default: "offers.csv"
+    description: "CSV with header row, columns: candidate_name,expiry_date(YYYY-MM-DD),status(pending/accepted/declined)."
+  warn_days:
+    default: 2
+    description: Days before offer expiry to start warning.
+compatible_agents: [generic, claude, codex]
+notes: Keep offers.csv updated as offers go out and candidates respond — one row per open offer.

package/jobs/hospitality/health-inspection-due-reminder.yaml

@@ -0,0 +1,32 @@
+id: health-inspection-due-reminder
+version: 1
+name: Health Inspection Due Reminder
+description: >
+  Warns you before a health department inspection is expected, based on
+  your jurisdiction's typical interval. Use this if an inspector has ever
+  shown up to a kitchen that wasn't ready because nobody was tracking when
+  the last one happened.
+category: hospitality
+tags: [hospitality, health-inspection, compliance]
+schedule: "0 8 1 * *"
+timezone: "UTC"
+runner: shell
+command: >
+  today=$(date +%s);
+  last_ts=$(date -d "{{last_inspection_date}}" +%s 2>/dev/null || date -jf "%Y-%m-%d" "{{last_inspection_date}}" +%s);
+  days_since=$(( (today - last_ts) / 86400 ));
+  days_left=$(( {{typical_interval_days}} - days_since ));
+  if [ "$days_left" -le "{{warn_days}}" ]; then echo "WARNING: next health inspection likely due within $days_left days (last was {{last_inspection_date}})"; exit 1;
+  else echo "OK: next inspection likely due in about $days_left days"; fi
+variables:
+  last_inspection_date:
+    default: "2026-01-01"
+    description: Date of the last health inspection (YYYY-MM-DD).
+  typical_interval_days:
+    default: 180
+    description: Typical number of days between inspections in your jurisdiction.
+  warn_days:
+    default: 14
+    description: Days before the estimated next inspection to start warning.
+compatible_agents: [generic, claude, codex]
+notes: This is an estimate based on your jurisdiction's typical cadence, not an actual scheduled date — most health departments don't publish inspection dates in advance.

package/jobs/hr/benefits-enrollment-deadline-reminder.yaml

@@ -0,0 +1,34 @@
+id: benefits-enrollment-deadline-reminder
+version: 1
+name: Benefits Open Enrollment Deadline Reminder
+description: >
+  Counts down to the open-enrollment deadline and flags employees who
+  haven't submitted elections yet. Use this if open enrollment has ever
+  closed with people accidentally defaulting to no coverage or last
+  year's plan.
+category: hr
+tags: [hr, benefits, enrollment, deadline]
+schedule: "0 9 * * *"
+timezone: "UTC"
+runner: agent-prompt
+prompt: |
+  Open enrollment closes on {{enrollment_deadline}}. Check
+  {{benefits_system}} for who has and hasn't submitted elections yet.
+  1. Report days remaining until the deadline.
+  2. List employees with no submitted election.
+  3. If within {{final_warn_days}} days of the deadline, draft an urgent
+     reminder message to the non-responders and, separately, a summary
+     for HR of who still needs a nudge.
+  4. If the deadline has passed, report final completion rate instead.
+variables:
+  enrollment_deadline:
+    default: "YYYY-MM-DD"
+    description: Date open enrollment closes.
+  benefits_system:
+    default: "your benefits administration platform"
+    description: Name of the benefits/HRIS system to check.
+  final_warn_days:
+    default: 5
+    description: Days before deadline to switch from routine reminders to urgent ones.
+compatible_agents: [claude, codex, hermes, openclaw, generic]
+notes: Draft-only — review reminder messages before sending, especially anything framed as "urgent" to employees.

package/jobs/hr/new-hire-paperwork-check.yaml

@@ -0,0 +1,35 @@
+id: new-hire-paperwork-check
+version: 1
+name: New Hire Paperwork Completion Check
+description: >
+  Checks which recently started employees still have incomplete
+  onboarding paperwork (I-9, tax forms, benefits enrollment, signed
+  handbook) and flags them before they become a compliance problem. Use
+  this if onboarding paperwork has ever slipped through the cracks past
+  the legal deadline.
+category: hr
+tags: [hr, onboarding, compliance, paperwork]
+schedule: "0 9 * * 1"
+timezone: "UTC"
+runner: agent-prompt
+prompt: |
+  Check onboarding status in {{hris_system}} for anyone who started in the
+  last {{lookback_days}} days.
+  1. For each new hire, list which required documents are still
+     incomplete: {{required_documents}}.
+  2. Flag anyone approaching or past a legal deadline (e.g. I-9 must be
+     completed within 3 business days of start in the US) as urgent.
+  3. Draft a reminder message to the employee and/or their manager for
+     each incomplete item.
+variables:
+  hris_system:
+    default: "your HRIS/onboarding system"
+    description: Name of the HR system to check (e.g. "BambooHR", "Rippling", "Gusto").
+  lookback_days:
+    default: 30
+    description: How far back to check for recent hires.
+  required_documents:
+    default: "I-9, W-4, direct deposit form, benefits enrollment, signed handbook acknowledgment"
+    description: Comma-separated list of documents your onboarding requires.
+compatible_agents: [claude, codex, hermes, openclaw, generic]
+notes: Legal deadlines for onboarding paperwork vary by country/jurisdiction — confirm the specifics for yours rather than relying on the example given.

package/jobs/hr/payroll-run-reminder.yaml

@@ -0,0 +1,30 @@
+id: payroll-run-reminder
+version: 1
+name: Payroll Run Cutoff Reminder
+description: >
+  Reminds you before the payroll submission cutoff so timesheets and
+  changes get in on time. Use this if payroll has ever run late because a
+  timesheet or new-hire update missed the cutoff by a day.
+category: hr
+tags: [payroll, hr, reminder, deadline]
+schedule: "0 9 * * *"
+timezone: "America/New_York"
+runner: shell
+command: >
+  today=$(date +%d);
+  cutoff="{{cutoff_day_of_month}}";
+  if [ "$today" -ge $((cutoff - {{warn_days_before}})) ] && [ "$today" -le "$cutoff" ]; then
+    echo "PAYROLL CUTOFF APPROACHING: due on day $cutoff of this month, today is day $today.";
+    echo "Checklist: timesheets submitted, new hires entered, terminations processed, PTO adjustments logged.";
+  else
+    echo "No payroll cutoff in the warning window today.";
+  fi
+variables:
+  cutoff_day_of_month:
+    default: 25
+    description: Day of the month payroll submissions are due.
+  warn_days_before:
+    default: 3
+    description: How many days before the cutoff to start reminding.
+compatible_agents: [generic, claude, codex]
+notes: Assumes a fixed monthly cutoff day — adjust the command or add a variable if your payroll provider uses a different schedule (biweekly, semi-monthly).

package/jobs/insurance/claim-status-check.yaml

@@ -0,0 +1,30 @@
+id: claim-status-check
+version: 1
+name: Open Claim Status Check
+description: >
+  Checks the status of any open insurance claims and flags ones that
+  haven't moved in a while. Use this if a claim has ever gone quiet for
+  weeks because nobody was following up with the adjuster.
+category: insurance
+tags: [insurance, claims, follow-up]
+schedule: "0 9 * * 1"
+timezone: "UTC"
+runner: agent-prompt
+prompt: |
+  Check the status of each open claim listed: {{open_claims}} (via the
+  insurer's portal, email correspondence, or whatever access is
+  available).
+  1. Report current status and last update date for each.
+  2. Flag any claim with no status change in more than
+     {{stale_after_days}} days as needing follow-up.
+  3. Draft a short follow-up message to the adjuster for each stale claim,
+     referencing the claim number and asking for a status update.
+variables:
+  open_claims:
+    default: "list your open claim numbers here"
+    description: Comma-separated list of open claim numbers/references to track.
+  stale_after_days:
+    default: 14
+    description: Days without a status update before a claim counts as stale.
+compatible_agents: [claude, codex, hermes, openclaw, generic]
+notes: Draft-only — review before sending anything to an adjuster or insurer.

package/jobs/insurance/coverage-gap-audit.yaml

@@ -0,0 +1,37 @@
+id: coverage-gap-audit
+version: 1
+name: Coverage Gap Audit
+description: >
+  Compares your current insurance policies against a standard coverage
+  checklist for your situation and flags anything missing or underinsured.
+  Use this if you've ever added a policy for one specific thing and never
+  circled back to check the overall picture still made sense.
+category: insurance
+tags: [insurance, coverage, audit, risk]
+schedule: "0 9 1 1,7 *"
+timezone: "UTC"
+runner: agent-prompt
+prompt: |
+  Review the current policies listed: {{current_policies}}, for a
+  {{profile_type}} profile.
+  1. Compare against a standard coverage checklist for that profile type
+     (e.g. for a homeowner: dwelling, liability, flood/earthquake if
+     applicable, umbrella; for a small business: general liability,
+     property, cyber, workers' comp as relevant).
+  2. Flag anything commonly recommended that's missing entirely.
+  3. Flag any policy whose coverage limit looks low relative to
+     {{asset_value_estimate}} in assets/exposure, with a one-line reason.
+  4. This is a gap-finding pass, not financial/legal advice — say so, and
+     recommend confirming any changes with a licensed agent.
+variables:
+  current_policies:
+    default: "list your current policies and coverage limits here"
+    description: Summary of currently held policies and their limits.
+  profile_type:
+    default: "homeowner"
+    description: Who this audit is for (e.g. "homeowner", "renter", "small business", "landlord").
+  asset_value_estimate:
+    default: "not specified"
+    description: Rough total value of assets/exposure to size coverage against.
+compatible_agents: [claude, codex, hermes, openclaw, generic]
+notes: General-knowledge gap-finding only — not a substitute for advice from a licensed insurance professional.

package/jobs/insurance/policy-renewal-reminder.yaml

@@ -0,0 +1,35 @@
+id: policy-renewal-reminder
+version: 1
+name: Policy Renewal Reminder
+description: >
+  Warns you before an insurance policy's renewal date so you have time to
+  shop around or review coverage, instead of auto-renewing on autopilot.
+  Use this if you've ever realized a policy renewed months ago at a higher
+  rate without you noticing.
+category: insurance
+tags: [insurance, renewal, policy, reminder]
+schedule: "0 8 * * *"
+timezone: "UTC"
+runner: shell
+command: >
+  awk -F',' -v days="{{warn_days_before}}" '
+    NR==1 {next}
+    {
+      cmd = "date -d \"" $2 "\" +%s 2>/dev/null || date -j -f %Y-%m-%d \"" $2 "\" +%s 2>/dev/null";
+      cmd | getline expiry_epoch; close(cmd);
+      "date +%s" | getline now_epoch; close("date +%s");
+      diff_days = (expiry_epoch - now_epoch) / 86400;
+      if (diff_days <= days && diff_days >= 0) {
+        printf "RENEWAL DUE: %s renews %s (%d days)\n", $1, $2, diff_days;
+      }
+    }
+  ' "{{policies_csv_path}}"
+variables:
+  policies_csv_path:
+    default: "./insurance-policies.csv"
+    description: "Path to a CSV with columns: policy_name,renewal_date (YYYY-MM-DD)."
+  warn_days_before:
+    default: 30
+    description: How many days before renewal to start warning.
+compatible_agents: [generic, claude, codex]
+notes: Maintain the CSV manually or export it from whatever policy-management tool/broker portal you use.

package/jobs/inventory/expiring-perishable-stock-check.yaml

@@ -0,0 +1,24 @@
+id: expiring-perishable-stock-check
+version: 1
+name: Expiring Perishable Stock Check
+description: >
+  Flags perishable inventory nearing its expiry date so it can be sold,
+  discounted, or used before it's wasted. Use this if stock has ever
+  expired on the shelf because nobody was tracking dates until it was too
+  late.
+category: inventory
+tags: [inventory, perishables, waste]
+schedule: "0 6 * * *"
+timezone: "UTC"
+runner: shell
+command: >
+  psql "{{database_url}}" -t -c "select sku, quantity, expiry_date from inventory where expiry_date is not null and expiry_date <= current_date + interval '{{warn_days}} days' order by expiry_date;"
+variables:
+  database_url:
+    default: "postgres://user:pass@host:5432/db"
+    description: Connection string for the inventory database.
+  warn_days:
+    default: 5
+    description: Days before expiry_date to start flagging stock.
+compatible_agents: [generic, claude, codex]
+notes: Assumes an `inventory` table with a nullable `expiry_date` column — adjust the query to your schema.

package/jobs/investing/dividend-payment-tracker.yaml

@@ -0,0 +1,34 @@
+id: dividend-payment-tracker
+version: 1
+name: Dividend Payment Tracker
+description: >
+  Checks that an expected dividend payment actually landed by its
+  scheduled pay date. Use this if a dividend has ever quietly not shown up
+  and it took weeks to notice something was wrong with a holding or
+  broker.
+category: investing
+tags: [investing, dividends, income]
+schedule: "0 9 * * *"
+timezone: "UTC"
+runner: shell
+command: >
+  today=$(date +%s);
+  awk -F, -v today="$today" '
+    NR>1 && $3 == "pending" {
+      cmd="date -d \"" $2 "\" +%s 2>/dev/null || date -jf \"%Y-%m-%d\" \"" $2 "\" +%s";
+      cmd | getline pay_ts; close(cmd);
+      if (pay_ts < today) print "WARNING: " $1 " dividend was due " $2 " and is still marked pending";
+    }
+  ' {{dividends_csv}}
+variables:
+  dividends_csv:
+    default: "dividends.csv"
+    description: "CSV with header row, columns: ticker,pay_date(YYYY-MM-DD),status(pending/received)."
+  warn_days:
+    default: 0
+    description: Unused placeholder kept for consistency with other date-check jobs; flags immediately once the pay date has passed.
+compatible_agents: [generic, claude, codex]
+notes: >
+  Update dividends.csv to "received" once a payment lands (check your
+  brokerage statement) — this job only flags entries still marked
+  "pending" past their pay date.

package/jobs/legal/nda-expiry-check.yaml

@@ -0,0 +1,31 @@
+id: nda-expiry-check
+version: 1
+name: NDA Expiry Check
+description: >
+  Warns you before a signed NDA reaches its expiration or renewal date.
+  Use this if an NDA has ever lapsed mid-relationship and nobody noticed
+  the confidentiality coverage had run out.
+category: legal
+tags: [legal, nda, contracts]
+schedule: "0 9 1 * *"
+timezone: "UTC"
+runner: shell
+command: >
+  today=$(date +%s);
+  awk -F, -v today="$today" -v warn={{warn_days}} '
+    NR>1 {
+      cmd="date -d \"" $2 "\" +%s 2>/dev/null || date -jf \"%Y-%m-%d\" \"" $2 "\" +%s";
+      cmd | getline exp_ts; close(cmd);
+      days_left = int((exp_ts - today) / 86400);
+      if (days_left <= warn) print "WARNING: NDA with \"" $1 "\" expires in " days_left " days (" $2 ")";
+    }
+  ' {{ndas_csv}}
+variables:
+  ndas_csv:
+    default: "ndas.csv"
+    description: "CSV with header row, columns: counterparty,expiry_date(YYYY-MM-DD)."
+  warn_days:
+    default: 30
+    description: Days before expiry to start warning.
+compatible_agents: [generic, claude, codex]
+notes: Keep ndas.csv updated whenever an NDA is signed — one row per counterparty/agreement.

package/jobs/logistics/freight-rate-spike-watch.yaml

@@ -0,0 +1,38 @@
+id: freight-rate-spike-watch
+version: 1
+name: Freight Rate Spike Watch
+description: >
+  Checks a freight lane's current quoted rate against your baseline and
+  flags a spike. Use this if a shipping rate has ever quietly doubled and
+  only got noticed when the invoice came in.
+category: logistics
+tags: [logistics, freight, rates]
+schedule: "0 7 * * 1"
+timezone: "UTC"
+runner: agent-prompt
+prompt: |
+  Check the current freight rate for the lane {{lane_description}} via
+  {{rate_source_hint}}.
+  1. Compare the current quote to the baseline rate of {{baseline_rate}}
+     per {{rate_unit}}.
+  2. Flag if it's {{spike_threshold_pct}}% or more above baseline.
+  3. Report the current rate, the delta, and if visible, the likely driver
+     (fuel surcharge, capacity crunch, seasonal demand).
+variables:
+  lane_description:
+    default: "Los Angeles, CA to Chicago, IL"
+    description: Origin/destination lane to check.
+  rate_source_hint:
+    default: "your freight broker or a rate index (DAT, Freightos)"
+    description: Where to check current freight rates.
+  baseline_rate:
+    default: 2.50
+    description: Baseline rate per unit to compare against.
+  rate_unit:
+    default: "mile"
+    description: Unit the rate is quoted in.
+  spike_threshold_pct:
+    default: 20
+    description: Percent above baseline that counts as a spike.
+compatible_agents: [claude, codex, hermes, openclaw, generic]
+notes: No generic script mode — freight rate sources vary (broker quotes, load boards, rate indices) and most don't expose a simple public API.

package/jobs/manufacturing/preventive-maintenance-due-check.yaml

@@ -0,0 +1,31 @@
+id: preventive-maintenance-due-check
+version: 1
+name: Preventive Maintenance Due Check
+description: >
+  Flags equipment coming due for scheduled preventive maintenance. Use
+  this if a machine has ever failed unexpectedly that would've been
+  caught by a PM visit nobody scheduled in time.
+category: manufacturing
+tags: [manufacturing, maintenance, equipment]
+schedule: "0 7 * * 1"
+timezone: "UTC"
+runner: shell
+command: >
+  today=$(date +%s);
+  awk -F, -v today="$today" -v warn={{warn_days}} '
+    NR>1 {
+      cmd="date -d \"" $2 "\" +%s 2>/dev/null || date -jf \"%Y-%m-%d\" \"" $2 "\" +%s";
+      cmd | getline due_ts; close(cmd);
+      days_left = int((due_ts - today) / 86400);
+      if (days_left <= warn) print "WARNING: equipment \"" $1 "\" PM due in " days_left " days (" $2 ")";
+    }
+  ' {{pm_schedule_csv}}
+variables:
+  pm_schedule_csv:
+    default: "pm-schedule.csv"
+    description: "CSV with header row, columns: equipment_id,next_pm_date(YYYY-MM-DD)."
+  warn_days:
+    default: 7
+    description: Days before the scheduled PM date to start warning.
+compatible_agents: [generic, claude, codex]
+notes: Keep pm-schedule.csv updated from your CMMS export or maintenance log before each run.

package/jobs/marketing/competitor-page-diff-watch.yaml

@@ -0,0 +1,48 @@
+id: competitor-page-diff-watch
+version: 1
+name: Competitor Page Diff Watch
+description: >
+  Snapshots a competitor's pricing or landing page and diffs it against the
+  last snapshot, turning any real change into a short action memo instead
+  of a wall of markup diff. Use this if a competitor has ever changed
+  pricing or repositioned a page and you found out from a customer instead
+  of noticing it yourself.
+category: marketing
+tags: [competitor, monitoring, pricing, marketing]
+schedule: "0 9 * * 1"
+timezone: "UTC"
+runner: hybrid
+command: >
+  snapshot_dir="{{snapshot_dir}}";
+  mkdir -p "$snapshot_dir";
+  today_file="$snapshot_dir/$(date +%Y-%m-%d).html";
+  curl -s -A "Mozilla/5.0" "{{competitor_url}}" -o "$today_file";
+  prev_file=$(ls -1 "$snapshot_dir" | grep -v "$(basename "$today_file")" | sort | tail -1);
+  if [ -n "$prev_file" ]; then
+    diff "$snapshot_dir/$prev_file" "$today_file" | head -100;
+  else
+    echo "first snapshot taken, nothing to diff against yet.";
+  fi
+prompt: |
+  Compare today's snapshot of {{competitor_url}} against the previous one
+  (raw HTML diff provided by the script, or fetch and compare yourself).
+  1. Ignore markup/whitespace/tracking-param noise.
+  2. Summarize any real content change: pricing, plan names, feature
+     claims, or messaging changes.
+  3. If nothing meaningful changed, say so in one line.
+  4. If something did change, draft a short internal action memo: what
+     changed, why it might matter, and one suggested response (match
+     pricing, update our comparison page, no action needed).
+script_note: >
+  `command` only saves a raw HTML diff — full of markup and whitespace
+  noise, no interpretation of what actually changed or whether it matters.
+  `prompt` filters the noise and produces an actionable summary.
+variables:
+  competitor_url:
+    default: "https://example.com/pricing"
+    description: Competitor page URL to monitor (pricing/landing pages work best).
+  snapshot_dir:
+    default: "./competitor-snapshots"
+    description: Local directory to store daily HTML snapshots for diffing.
+compatible_agents: [claude, codex, hermes, openclaw, generic]
+notes: Respect the target site's robots.txt/terms before scraping; some sites render pricing via JS and won't diff meaningfully from raw HTML alone.