Add comprehensive safety and robustness guidelines to system prompt

- Add data validation checks (empty dataframes, missing values)
- Add error handling with try-except blocks
- Add city/location validation before filtering
- Add proper handling of empty results after filtering
- Add numerical formatting (.round(2)) to avoid long decimals
- Add division by zero protection
- Add date range validation
- Add proper units formatting (μg/m³)
- Add memory management (plt.close())
- Add column name validation
- This should make the generated code much more robust and safe

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

src.py

@@ -295,12 +295,25 @@ WHEN TO CREATE PLOTS vs TEXT ANSWERS:
 - Questions asking for comparisons of many items → PLOTS
 - Simple direct questions → TEXT ANSWERS
 
-Requirements:
+SAFETY & ROBUSTNESS RULES:
+- Always check if data exists before processing: if df.empty: answer = "No data available"
+- Handle missing values: use .dropna() or .fillna() appropriately
+- Use try-except blocks for risky operations like indexing
+- Validate city/location names exist in data before filtering
+- Check for empty results after filtering: if filtered_df.empty: answer = "No data found for specified criteria"
+- Use .round(2) for numerical results to avoid long decimals
+- Handle division by zero: check denominators before division
+- Validate date ranges exist in data
+- Use proper string formatting for answers with units (μg/m³)
+
+TECHNICAL REQUIREMENTS:
 - Save final result in variable called 'answer'
 - For TEXT: Store the direct answer as a string in 'answer'
 - For PLOTS: Save with unique filename f"plot_{{uuid.uuid4().hex[:8]}}.png" and store filename in 'answer'
 - Convert numpy types to int when using as indices: int(value)
 - Always use .iloc or .loc properly for pandas indexing
+- Close matplotlib figures with plt.close() to prevent memory leaks
+- Use proper column name checks before accessing columns
 """
 
         query = f"""{system_prompt}