Code Interpreter (Data Insights)

In this level, you’ll give your agent analytical superpowers.
Your agent will load sales data and generate graphs and summaries—useful for managers, dashboards, and feature decisions.
Think: “Show sales by store and top toppings last week” → the agent produces charts and a short write-up. 📊🍕

📋 Tasks

• Enable Code Interpreter (or an equivalent sandboxed execution tool) for your agent.
• Provide the dataset from data/sales-data/ to the runtime.
  • Download the dataset
• Ask the agent to create graphical reports (e.g., sales by store, top toppings, hourly demand).
• Have the agent export charts (PNG/SVG) and a brief insights summary (bulleted or paragraph).
• (Optional) Save outputs to a reports/ folder and list file paths in the response.

✅ Pass Criteria and Requirements

• The agent loads the sales data.
• The agent produces at least two charts (e.g., bar/line/pie) and a concise analysis describing trends or outliers.
• Outputs are accessible (file paths or attachments) and reproducible.

🛠️ Hints & Tips

• Prompt the agent to explain assumptions (date parsing, missing values, currency).
• Encourage clean column naming and data validation (row counts, null checks) before plotting.
• Prefer deterministic outputs: fix random seeds, specify chart titles/axes/units.
• If data volume is large, ask the agent to sample or aggregate before plotting.

📚 Resources

• Your project’s data/sales-data/ folder (sample CSV/Parquet files)
• Visualization prompts:
  • “Create a bar chart of total sales by store.”
  • “Top 5 toppings by revenue last month.”
  • “Hourly order volume trend for the last 7 days.”