1. Data Generation (mcp/data))
-
create_db.py
Generates the entire customer, subscription, billing, and usage database (contoso.db) with:- 250 randomized customers
- 9 deterministic “AI challenge” scenarios
- Linked tables for payments, promotions, support, incidents, etc.
- Scenario answer keys written out to
customer_scenarios.md - Optionally creates knowledge document embeddings if Azure OpenAI is configured
-
contoso.db
The generated SQLite database (runpython data/create_db.pyto generate). -
SCENARIOS.md
All nine "challenge" scenarios (e.g. “invoice anomaly”, “promotion eligibility”), each with:- Customer context
- Challenge prompt
- Detailed solution (“answer-key” for agents)
-
kb.json
Knowledge base entries used for agent tool lookups, including policies, procedures, troubleshooting guides, etc.