dojo.md 0.1.0

package/courses/ecommerce-support-tickets/scenarios/level-3/ecommerce-support-shift-advanced.yaml

@@ -0,0 +1,79 @@
+meta:
+  id: ecommerce-support-shift-advanced
+  level: 3
+  course: ecommerce-support-tickets
+  type: output
+  description: "Navigate an advanced e-commerce support shift — handle a complex queue with VIP customers, fraud alerts, international issues, and team coordination"
+  tags: [support, ecommerce, simulation, advanced-shift, complex, advanced]
+
+state: {}
+
+trigger: |
+  An advanced e-commerce support shift includes everything from
+  the basics plus: VIP customer management, fraud detection,
+  international complications, team coordination, and real-time
+  decision-making under pressure.
+
+  You're the senior agent on a Friday afternoon shift at
+  "LuxeBeauty" — a premium cosmetics e-commerce brand.
+  Your team: You + 2 agents (Maya and Jake).
+  It's December 15th — peak holiday season.
+
+  HOUR 1 (1:00 PM - 2:00 PM):
+  Queue:
+  - Chat 1: VIP customer ($3,000/year) — "My holiday gift set
+    order shows 'processing' for 3 days. I need it by the 20th
+    for a party. This is a $250 order."
+  - Email 1: International (UK) — return request for items
+    totaling £180, outside your normal 30-day window but within
+    UK's 14-day cooling-off period
+  - Email 2: Fraud alert — a $1,200 order flagged by your
+    system (5 luxury gift sets to a new address with new account)
+
+  At 1:30 PM: Maya asks for help — a customer is claiming they
+  never received a package (tracking shows delivered with photo
+  proof of delivery at correct doorstep)
+
+  HOUR 2 (2:00 PM - 3:00 PM):
+  - Instagram DM: Customer posts story showing a broken compact
+    mirror from their order, tags the brand with angry caption
+    (500 views and growing)
+  - Chat 2: Subscription customer — wants to pause but is also
+    upset about being charged for a box she thought she cancelled
+  - Email 3: "I placed 2 orders yesterday by accident. Can you
+    combine them to avoid double shipping charges?"
+
+  At 2:45 PM: Jake reports 6 similar complaints about a new
+  eye shadow palette — appears to be a product quality batch issue
+
+  HOUR 3 (3:00 PM - 4:00 PM):
+  - The product quality issue escalates (15 complaints now)
+  - A beauty blogger (100K followers) DMs saying she received a
+    damaged PR package and wants a replacement before her review
+    goes live Monday
+  - Email 4: B2B inquiry from a spa asking about wholesale pricing
+
+  For the entire shift:
+  1. Prioritize and sequence your actions for each hour
+  2. Write key responses (at least 8 complete responses)
+  3. Handle team coordination
+  4. Make decisions on the fraud alert and product quality issue
+  5. Write end-of-shift handoff notes
+
+assertions:
+  - type: llm_judge
+    criteria: "Priority decisions are sound — the VIP holiday order, fraud alert, and emerging product quality issue get appropriate urgency"
+    weight: 0.30
+    description: "Sound priority decisions"
+  - type: llm_judge
+    criteria: "The product quality batch issue is escalated properly — 15 complaints about the same product triggers an internal quality alert"
+    weight: 0.25
+    description: "Proper quality issue escalation"
+  - type: llm_judge
+    criteria: "The beauty blogger PR package gets strategic treatment — it's handled as both a support issue and a marketing opportunity"
+    weight: 0.25
+    description: "Strategic PR package handling"
+  - type: llm_judge
+    criteria: "Team coordination is effective — Maya and Jake get clear guidance and the workload is distributed appropriately"
+    weight: 0.20
+    description: "Effective team coordination"

package/courses/ecommerce-support-tickets/scenarios/level-3/ecommerce-support-team-building.yaml

@@ -0,0 +1,119 @@
+meta:
+  id: ecommerce-support-team-building
+  level: 3
+  course: ecommerce-support-tickets
+  type: output
+  description: "Build an e-commerce support team — hire, train, and develop agents with product knowledge, sales skills, and brand voice mastery"
+  tags: [support, ecommerce, team-building, hiring, training, advanced]
+
+state: {}
+
+trigger: |
+  E-commerce support agents need a unique blend of skills:
+  customer service empathy, product knowledge, sales instinct,
+  brand voice consistency, and technical tool proficiency.
+  Building a great e-commerce support team requires hiring for
+  these traits and developing them through structured training.
+
+  Part 1: E-commerce agent skills
+  1. Essential skills:
+     - Product knowledge (your entire catalog, not just FAQs)
+     - Brand voice mastery (consistent tone across all channels)
+     - Sales instinct (when to recommend, cross-sell, upsell)
+     - Technical proficiency (help desk, order system, analytics)
+     - Emotional intelligence (reading customer mood, de-escalation)
+     - Speed (e-commerce customers expect fast resolution)
+  2. Hiring for e-commerce:
+     - Retail experience (understanding customer buying behavior)
+     - Writing quality (most support is text-based)
+     - Product passion (they should like what you sell)
+     - Multi-tasking ability (chat + email simultaneously)
+     - Sales background is a bonus (not required)
+  3. Training unique to e-commerce:
+     - Product training (hands-on with every product category)
+     - Order management system training
+     - Returns and refund processing
+     - Fraud detection basics
+     - Brand voice certification
+     - Sales techniques appropriate for support
+
+  Part 2: Build the team
+  For "ActiveWear" — an athletic apparel e-commerce brand
+  (growing from 3 to 8 support agents over 6 months):
+
+  1. Create the job description:
+     Write a compelling job posting for an e-commerce support
+     specialist that attracts candidates who are:
+     - Customer-focused with retail or service experience
+     - Strong writers who can match brand voice
+     - Comfortable with technology and multi-channel work
+     - Passionate about fitness and active lifestyles
+
+  2. Design the interview process:
+     Step 1 — Written assessment:
+     3 simulated customer scenarios:
+     a) WISMO ticket (basic, tests process knowledge)
+     b) Product recommendation chat (tests product knowledge and
+        sales instinct)
+     c) Angry customer return request (tests empathy and de-escalation)
+     Include scoring rubric.
+
+     Step 2 — Live role play:
+     15-minute simulation where candidate handles a live chat
+     while simultaneously drafting an email response.
+     Include evaluation criteria.
+
+  3. Create the 30-day onboarding program:
+     Week 1: Brand, product, and culture immersion
+     - Product knowledge training (hands-on with merchandise)
+     - Brand voice workshop (writing exercises)
+     - Tool training (help desk, order system, shipping)
+     - Shadow experienced agents
+
+     Week 2: Process and policy mastery
+     - Return/refund processing (practice scenarios)
+     - Fraud detection basics
+     - Escalation procedures
+     - Channel-specific guidelines (email, chat, social)
+
+     Week 3: Supervised practice
+     - Handle tickets with mentor review
+     - Chat under supervision
+     - Social media responses (drafted, reviewed, then posted)
+
+     Week 4: Graduated independence
+     - Independent email handling
+     - Supervised chat (mentor available)
+     - First QA review and coaching
+     - 30-day assessment and goal setting
+
+  4. Design ongoing development:
+     - Monthly product knowledge updates (new arrivals)
+     - Quarterly brand voice refreshers
+     - Sales skills workshops
+     - Career path: Agent → Senior → Lead → Manager
+     - Cross-training opportunities (merchandising, marketing)
+
+  Part 3: Team culture
+  1. Building a team that cares about the brand
+  2. Incentive structures for e-commerce support
+  3. Preventing burnout in seasonal peaks
+  4. Support team as brand ambassadors
+
+assertions:
+  - type: llm_judge
+    criteria: "The job description is compelling and brand-aligned — it attracts candidates who are passionate about the product category, not just looking for any support job"
+    weight: 0.25
+    description: "Compelling brand-aligned job posting"
+  - type: llm_judge
+    criteria: "The interview process evaluates e-commerce-specific skills — writing quality, product knowledge, and multi-tasking ability are tested"
+    weight: 0.30
+    description: "E-commerce skill evaluation"
+  - type: llm_judge
+    criteria: "The 30-day onboarding builds competence progressively — from observation to supervised practice to graduated independence at an appropriate pace"
+    weight: 0.25
+    description: "Progressive competence building"
+  - type: llm_judge
+    criteria: "Ongoing development includes product knowledge — new arrivals training reflects the reality that e-commerce catalogs constantly change"
+    weight: 0.20
+    description: "Continuous product knowledge"

package/courses/ecommerce-support-tickets/scenarios/level-3/ecommerce-voc-analytics.yaml

@@ -0,0 +1,82 @@
+meta:
+  id: ecommerce-voc-analytics
+  level: 3
+  course: ecommerce-support-tickets
+  type: output
+  description: "Analyze voice of customer data from e-commerce support — extract insights from tickets, reviews, chat logs, and social mentions to drive product and experience improvements"
+  tags: [support, ecommerce, voc, analytics, customer-feedback, advanced]
+
+state: {}
+
+trigger: |
+  Voice of Customer (VoC) analytics transforms raw support
+  interactions into strategic insights. E-commerce support teams
+  sit on a goldmine of customer feedback — every ticket, chat,
+  review, and social mention contains signals about product quality,
+  website experience, and unmet customer needs.
+
+  Part 1: VoC fundamentals
+  1. Sources of customer voice in e-commerce
+  2. Qualitative vs. quantitative feedback analysis
+  3. Turning unstructured feedback into structured data
+  4. Feedback loop: insight → action → measurement
+
+  Part 2: Analyze VoC data
+  For "StyleHub" — an online fashion retailer ($15M revenue,
+  60K customers, 8,000 tickets/month):
+
+  Recent VoC data snapshot:
+  - Top ticket themes: sizing issues (22%), late delivery (18%),
+    fabric quality (15%), wrong item received (12%), returns
+    process confusion (10%), payment failures (8%), other (15%)
+  - Product reviews (last 90 days): 4,200 reviews, avg 3.6 stars
+  - Common review phrases: "runs small," "color different from
+    photo," "great style but poor stitching," "fast shipping"
+  - Social mentions: 340/month, 45% negative sentiment
+  - NPS by segment: new customers 18, repeat customers 42
+  - Chat satisfaction: 78% (down from 85% three months ago)
+
+  1. Perform a comprehensive VoC analysis:
+     - Categorize feedback by theme, severity, and frequency
+     - Identify the top 5 actionable insights
+     - Calculate estimated revenue impact of each issue
+     - Map insights to responsible teams (product, logistics,
+       marketing, support, engineering)
+
+  2. Create a VoC reporting framework:
+     - Weekly VoC digest for support managers
+     - Monthly VoC report for cross-functional leadership
+     - Quarterly VoC strategic review for executives
+     - Real-time VoC alerts for critical emerging issues
+     Define content, audience, format, and cadence for each.
+
+  3. Design a closed-loop VoC program:
+     - How feedback is collected and tagged at source
+     - How insights are prioritized and routed
+     - How actions are tracked to completion
+     - How impact is measured and reported back
+     - How customers are informed their feedback led to changes
+
+  Part 3: Advanced VoC techniques
+  1. Sentiment analysis and trend detection
+  2. Competitive VoC benchmarking
+  3. Predictive analytics from VoC signals
+  4. Building a customer-centric culture through VoC
+
+assertions:
+  - type: llm_judge
+    criteria: "The VoC analysis correctly identifies actionable insights — sizing issues, photo accuracy, and quality problems are connected to both returns and negative reviews"
+    weight: 0.30
+    description: "Actionable VoC insights"
+  - type: llm_judge
+    criteria: "The reporting framework serves different audiences appropriately — weekly tactical for managers vs. quarterly strategic for executives"
+    weight: 0.25
+    description: "Audience-appropriate reporting"
+  - type: llm_judge
+    criteria: "Revenue impact estimates are reasonable — they connect customer issues to measurable business outcomes like return costs and churn"
+    weight: 0.25
+    description: "Revenue impact quantification"
+  - type: llm_judge
+    criteria: "The closed-loop program ensures feedback leads to action — there's a clear path from customer complaint to product improvement to customer notification"
+    weight: 0.20
+    description: "Closed-loop feedback program"

package/courses/ecommerce-support-tickets/scenarios/level-3/international-ecommerce-support.yaml

@@ -0,0 +1,109 @@
+meta:
+  id: international-ecommerce-support
+  level: 3
+  course: ecommerce-support-tickets
+  type: output
+  description: "Support international e-commerce customers — handle cross-border shipping, customs, duties, currency issues, and cultural differences in global support"
+  tags: [support, ecommerce, international, cross-border, customs, advanced]
+
+state: {}
+
+trigger: |
+  International e-commerce adds layers of complexity: customs
+  duties, longer shipping times, currency conversion, language
+  barriers, and different consumer protection laws by country.
+  Support agents need to navigate these while maintaining the
+  same friendly, helpful experience domestic customers receive.
+
+  Part 1: International support challenges
+  1. Shipping and logistics:
+     - Longer delivery times (7-21 days standard)
+     - Customs clearance delays
+     - Duties and taxes (DDP vs. DDU shipping)
+     - Carrier limitations by country
+     - Address format differences
+     - Restricted/prohibited items by country
+  2. Financial:
+     - Currency conversion and pricing
+     - International payment methods (iDEAL, Boleto, WeChat Pay)
+     - VAT/GST handling
+     - Refund in original currency
+     - Exchange rate fluctuation disputes
+  3. Legal and regulatory:
+     - EU consumer protection (14-day cooling-off period)
+     - GDPR data handling
+     - Product safety standards by region
+     - Import restrictions (cosmetics, food, electronics)
+     - Distance selling regulations
+
+  Part 2: Handle international support scenarios
+  For "GlobalStyle" — a US-based fashion brand shipping to 40 countries:
+
+  1. Customs duty surprise:
+     "I ordered a coat for €120 but was charged €35 in customs
+     duties on delivery! Your website didn't mention anything
+     about extra charges. I refuse to pay and want a refund."
+     Customer: Germany, first international order
+     - Explain duties and how they work
+     - Address the transparency issue
+     - Offer resolution (refund duties? absorb cost? DDP solution?)
+
+  2. International return logistics:
+     "I need to return a dress from Australia. The return shipping
+     label you sent is for USPS — it doesn't work here. And the
+     return shipping quote from Australia Post is $45 AUD!"
+     Customer: Australia, $89 USD order
+     - Handle the expensive international return
+     - Offer alternatives (partial refund, keep the item, local return)
+     - Review policy for international returns
+
+  3. Currency and pricing dispute:
+     "Your website showed the price as ¥12,000 but my credit card
+     was charged ¥12,847. That's a 7% difference! I want the
+     price I was shown."
+     Customer: Japan, credit card currency conversion
+     - Explain dynamic currency conversion vs. bank conversion
+     - Determine if the difference is fair
+     - Offer resolution if it's a pricing display issue
+
+  4. EU consumer rights (14-day cooling-off):
+     "Under EU law, I have the right to return this item within
+     14 days without giving a reason. You're telling me I need a
+     reason? That's illegal."
+     Customer: France, returning an unworn dress within 14 days
+     - Address the legal requirement correctly
+     - Process the return per EU regulations
+     - Ensure your policy is compliant
+
+  5. International shipping tracking gap:
+     "Tracking shows my package left the US 12 days ago but then
+     nothing. No updates from any carrier. Did it get lost in
+     the ocean?"
+     Customer: Brazil, order value $150
+     - Explain international tracking handoff (carrier to carrier)
+     - Investigate the gap
+     - Set realistic expectations
+
+  Part 3: International support operations
+  1. Building international support capabilities (language, timezone)
+  2. DDP vs. DDU shipping strategy
+  3. International return policy design
+  4. Country-specific support playbooks
+
+assertions:
+  - type: llm_judge
+    criteria: "Customs and duty explanations are accurate and clear — the customer understands why they were charged and what can be done about it"
+    weight: 0.30
+    description: "Accurate customs explanations"
+  - type: llm_judge
+    criteria: "EU consumer rights are handled correctly — the 14-day cooling-off period is recognized as a legal requirement, not a discretionary policy"
+    weight: 0.25
+    description: "Correct EU rights handling"
+  - type: llm_judge
+    criteria: "International returns offer practical alternatives — expensive cross-border shipping is addressed with creative solutions"
+    weight: 0.25
+    description: "Practical international return solutions"
+  - type: llm_judge
+    criteria: "Currency issues are explained without condescension — exchange rate mechanics are clarified in accessible language"
+    weight: 0.20
+    description: "Accessible currency explanations"

package/courses/ecommerce-support-tickets/scenarios/level-3/product-knowledge-management.yaml

@@ -0,0 +1,97 @@
+meta:
+  id: product-knowledge-management
+  level: 3
+  course: ecommerce-support-tickets
+  type: output
+  description: "Build product knowledge systems — create internal wikis, training materials, and quick-reference tools so agents can answer any product question instantly"
+  tags: [support, ecommerce, product-knowledge, wiki, training, advanced]
+
+state: {}
+
+trigger: |
+  E-commerce support agents are expected to know everything about
+  every product in the catalog. With catalogs ranging from hundreds
+  to tens of thousands of SKUs, this requires structured knowledge
+  management — not just memory. Great product knowledge systems
+  give agents instant access to accurate information.
+
+  Part 1: Product knowledge challenges
+  1. Scale challenges:
+     - Large catalogs (hundreds to thousands of products)
+     - Frequent changes (new arrivals, discontinued items)
+     - Seasonal inventory (holiday items, limited editions)
+     - Variant complexity (sizes, colors, materials, configurations)
+  2. Knowledge types needed:
+     - Specifications (dimensions, materials, weight, compatibility)
+     - Usage guidance (care instructions, best practices, limitations)
+     - Comparison data (how products differ from each other)
+     - Common issues (known defects, fit guidance, FAQ per product)
+     - Inventory status (in stock, backordered, discontinued)
+     - Promotional context (current sales, upcoming launches)
+  3. Knowledge delivery methods:
+     - Internal wiki (searchable, always current)
+     - Product quick-reference cards (one-pagers per product)
+     - Comparison matrices (side-by-side product comparisons)
+     - Known issue alerts (recently discovered problems)
+     - New arrival briefings (weekly/monthly knowledge updates)
+
+  Part 2: Build a product knowledge system
+  For "CycleGear" — an online cycling equipment store with
+  2,000 SKUs across bikes, components, clothing, and accessories:
+
+  1. Design the internal product wiki structure:
+     - Category organization (bikes, components, clothing, accessories)
+     - Product page template (what information every product entry needs)
+     - Quick search and cross-reference system
+     - Version control (who updated, when, what changed)
+     - Integration with help desk (inline product lookup)
+
+  2. Create 3 sample product knowledge pages:
+     a) "CycleGear Pro Road Bike" ($1,899):
+        Full specs, sizing guide, common questions, known issues,
+        comparison with similar models, care instructions
+     b) "Shimano 105 Groupset" ($699):
+        Compatibility matrix, installation notes, common issues,
+        comparison with Ultegra and Tiagra
+     c) "CycleGear Waterproof Jersey" ($89):
+        Material specs, size chart, care instructions, customer
+        feedback summary, comparison with non-waterproof version
+
+  3. Design the weekly knowledge update process:
+     - New arrivals briefing (Monday morning)
+     - Known issue alerts (as discovered, real-time)
+     - Price change notifications
+     - Seasonal inventory updates
+     - Product training sessions (monthly deep dives)
+
+  4. Create a product comparison tool:
+     Design a template that agents can quickly fill out when
+     customers ask "What's the difference between X and Y?"
+     - Feature comparison matrix
+     - Use case guidance (which is better for what scenario)
+     - Price/value positioning
+     - Customer review summary per product
+
+  Part 3: Knowledge maintenance
+  1. Keeping knowledge current (process for updates)
+  2. Product knowledge testing for agents
+  3. Using support data to improve product information
+  4. Measuring the impact of better product knowledge on metrics
+
+assertions:
+  - type: llm_judge
+    criteria: "The wiki structure is practical for 2,000 SKUs — it's searchable, well-organized, and agents can find information quickly during live conversations"
+    weight: 0.30
+    description: "Practical searchable wiki structure"
+  - type: llm_judge
+    criteria: "Sample product pages contain all information an agent needs — specs, sizing, known issues, and comparisons are all included"
+    weight: 0.25
+    description: "Complete product knowledge pages"
+  - type: llm_judge
+    criteria: "The weekly update process keeps knowledge current — new arrivals, price changes, and known issues are communicated systematically"
+    weight: 0.25
+    description: "Systematic knowledge updates"
+  - type: llm_judge
+    criteria: "The comparison tool helps agents make recommendations — it guides toward the right product for the customer's specific needs"
+    weight: 0.20
+    description: "Recommendation-enabling comparisons"

package/courses/ecommerce-support-tickets/scenarios/level-3/supply-chain-support.yaml

@@ -0,0 +1,101 @@
+meta:
+  id: supply-chain-support
+  level: 3
+  course: ecommerce-support-tickets
+  type: output
+  description: "Support customers through supply chain disruptions — communicate backorders, delays, substitutions, and stock-outs while maintaining trust"
+  tags: [support, ecommerce, supply-chain, backorders, inventory, advanced]
+
+state: {}
+
+trigger: |
+  Supply chain disruptions directly impact customer experience.
+  When products are backordered, shipments delayed, or items
+  out of stock after ordering, support must communicate honestly,
+  offer alternatives, and maintain customer trust through
+  uncertainty.
+
+  Part 1: Supply chain impacts on support
+  1. Common supply chain issues:
+     - Product out of stock after order placed
+     - Backorder with uncertain delivery date
+     - Supplier quality issues (defective batch)
+     - Shipping carrier disruptions (weather, strikes, capacity)
+     - Warehouse capacity issues (fulfillment delays)
+     - Global supply chain disruptions (port congestion, shortages)
+  2. Communication challenges:
+     - Delivering bad news (your order is delayed/cancelled)
+     - Managing uncertain timelines (we don't know when)
+     - Offering substitutions (similar but not identical)
+     - Proactive vs. reactive communication
+     - Balancing honesty with maintaining confidence
+
+  Part 2: Handle supply chain support scenarios
+  For "OutdoorEdge" — an outdoor recreation e-commerce store:
+
+  1. Backorder with uncertain timeline:
+     200 customers ordered the popular "Summit Pro" tent that went
+     out of stock mid-order rush. Supplier says 4-8 weeks for restock
+     but can't guarantee the date.
+     - Write the proactive notification to all 200 customers
+     - Offer options: wait, substitute, cancel with bonus
+     - Handle a customer who replies angrily
+
+  2. Partial order cancellation (item discontinued):
+     A customer ordered 5 items. One item (a specific hiking pole
+     model) has been discontinued by the manufacturer with no
+     replacement. The other 4 items are ready to ship.
+     - Communicate the cancellation of 1 item
+     - Ship the remaining 4 without delay
+     - Suggest an alternative product
+     - Refund the cancelled item
+
+  3. Quality issue requiring hold on shipments:
+     Your QA team discovers that a batch of water filters may not
+     meet filtration standards. You have 85 orders containing these
+     filters ready to ship or already shipped.
+     - Stop unshipped orders (communicate the delay)
+     - Recall already-shipped orders (safety concern)
+     - Explain without causing panic (water safety issue)
+
+  4. Holiday season supplier delay:
+     It's December 5th. Your top-selling gift items (premium thermos
+     set, $89) are stuck at the port. They won't arrive at your
+     warehouse until December 18th. You have 500 pre-orders
+     promising delivery by Christmas.
+     - Calculate if Christmas delivery is still possible
+     - Write communication for different scenarios (possible vs. not)
+     - Offer alternatives for those who can't wait
+     - Manage the gift-giving anxiety
+
+  5. Substitution offer:
+     A customer ordered specific running shoes (Model X, $129) that
+     are now permanently out of stock. The closest alternative is
+     Model Y ($149, slightly different features).
+     - Offer the substitution with full transparency
+     - Absorb the price difference (don't charge more)
+     - Let them decide (substitution or refund)
+
+  Part 3: Supply chain resilience
+  1. Building support playbooks for common supply chain scenarios
+  2. Proactive communication templates
+  3. Working with operations on supply chain visibility
+  4. Customer expectation management during disruptions
+
+assertions:
+  - type: llm_judge
+    criteria: "All communications are honest about uncertainty — when delivery dates are unknown, the response acknowledges this rather than inventing dates"
+    weight: 0.30
+    description: "Honest uncertainty communication"
+  - type: llm_judge
+    criteria: "Customer options are clearly presented — each scenario gives customers clear choices (wait, alternative, refund) with enough information to decide"
+    weight: 0.25
+    description: "Clear customer options"
+  - type: llm_judge
+    criteria: "The safety concern (water filters) is handled with appropriate urgency — product safety outweighs business convenience"
+    weight: 0.25
+    description: "Appropriate safety urgency"
+  - type: llm_judge
+    criteria: "The holiday delivery scenario addresses gift anxiety — the emotional dimension of Christmas gift expectations is acknowledged"
+    weight: 0.20
+    description: "Gift anxiety acknowledgment"

package/courses/ecommerce-support-tickets/scenarios/level-4/complex-ecommerce-simulation.yaml

@@ -0,0 +1,80 @@
+meta:
+  id: complex-ecommerce-simulation
+  level: 4
+  course: ecommerce-support-tickets
+  type: output
+  description: "Navigate a complex multi-day e-commerce scenario — manage a holiday crisis, product recall, team restructuring, and strategic initiative simultaneously"
+  tags: [support, ecommerce, simulation, complex, multi-thread, expert]
+
+state: {}
+
+trigger: |
+  You are the Director of Customer Experience at "FreshMart" —
+  a premium grocery delivery e-commerce company
+  ($40M revenue, 150K customers, 35-person support team).
+
+  Navigate a critical week during the November holiday season:
+
+  MONDAY: Holiday prep week
+  Thread 1 — Volume forecast crisis:
+  Your data team projects 4x normal support volume for Thanksgiving
+  week (next week). You currently have capacity for 2x.
+  Task: Create an emergency staffing and automation plan for
+  Thanksgiving week.
+
+  Thread 2 — Product quality alert:
+  Your food safety team reports that a batch of organic turkeys
+  from one supplier may have been stored at incorrect temperature.
+  2,000 turkeys from this batch are scheduled for delivery
+  Wednesday-Friday this week.
+  Task: Write the food safety response plan (this is a potential
+  health risk before a holiday where people are serving turkey).
+
+  WEDNESDAY:
+  Thread 1 update: You've secured 15 temporary agents but they
+  need training by Friday.
+  Task: Create a crash 2-day training program focused on holiday-
+  specific scenarios.
+
+  Thread 2 update: Lab results confirm 500 of the 2,000 turkeys
+  are from the affected batch. 200 have already been delivered.
+  Task: Write urgent customer notification and initiate replacement
+  deliveries before Thanksgiving.
+
+  Thread 3 — Competitor promotion:
+  Your main competitor launches "Free Thanksgiving Delivery" with
+  massive social media campaign. Your inbox fills with "Why don't
+  you offer free delivery?" complaints. 50 cancellation requests.
+  Task: Write the competitive response strategy and cancellation
+  save approach.
+
+  FRIDAY:
+  Thread 2 resolution: All affected customers contacted, replacement
+  turkeys delivered. One customer consumed the turkey and is
+  reporting mild food poisoning symptoms.
+  Task: Handle the food illness report (legal, safety, customer care).
+
+  Thread 4 — Strategic initiative:
+  CEO wants you to present a plan for launching a customer loyalty
+  program that integrates with support. Board meeting in 2 weeks.
+  Task: Write the loyalty program concept including support integration.
+
+  End of week: Write your weekly report covering all 4 threads.
+
+assertions:
+  - type: llm_judge
+    criteria: "The food safety response prioritizes customer health — turkey replacement is executed urgently before Thanksgiving without waiting for full investigation"
+    weight: 0.30
+    description: "Health-first food safety response"
+  - type: llm_judge
+    criteria: "The crash training program is realistic for 2 days — it focuses on the most critical holiday scenarios rather than trying to cover everything"
+    weight: 0.25
+    description: "Realistic crash training"
+  - type: llm_judge
+    criteria: "The food illness report is handled with appropriate legal and safety awareness — empathy without liability, internal escalation to legal"
+    weight: 0.25
+    description: "Legally aware illness response"
+  - type: llm_judge
+    criteria: "The weekly report synthesizes 4 complex threads coherently — it presents a clear picture of status, risks, and decisions needed"
+    weight: 0.20
+    description: "Coherent multi-thread synthesis"