improve data-analysis skill structure + description

- expand description with trigger terms for agent selection
- add structured workflow (load, profile, transform, visualize, report)
- add error recovery table for common data issues
- replace fragmented code snippets with end-to-end EDA example

Author: fernandezbaptiste

examples/agents/skills/data-analysis/SKILL.md

@@ -1,29 +1,58 @@
 ---
 name: data-analysis
-description: Analyze datasets, generate charts, and create summary reports. Use when the user needs to work with CSV, Excel, or other tabular data formats for analysis or visualization.
+description: >
+  Analyze datasets, generate charts, and create summary reports from CSV, Excel,
+  JSON, Parquet, or other tabular data. Capabilities: statistical profiling,
+  outlier detection, pivot tables, groupby aggregation, time-series analysis,
+  correlation matrices, and publication-ready visualizations.
+  Trigger terms: analyze data, plot chart, summarize CSV, data profiling,
+  statistics, histogram, scatter plot, dashboard, EDA, exploratory analysis.
 ---
 
 # Data Analysis
 
 ## When to use this skill
 Use this skill when the user needs to:
-- Analyze CSV or Excel files
-- Generate charts and visualizations
-- Calculate statistics and summaries
-- Clean and transform data
-
-## How to analyze data
-1. Use pandas for data analysis:
-   ```python
-   import pandas as pd
-   df = pd.read_csv('data.csv')
-   summary = df.describe()
-   ```
-
-## How to create visualizations
-1. Use matplotlib or seaborn for charts:
-   ```python
-   import matplotlib.pyplot as plt
-   df.plot(kind='bar')
-   plt.savefig('chart.png')
-   ```
+- Analyze CSV, Excel, JSON, or Parquet files
+- Generate charts and visualizations (bar, line, scatter, heatmap)
+- Calculate statistics, correlations, or distributions
+- Clean, transform, pivot, or aggregate data
+- Perform exploratory data analysis (EDA)
+
+## Workflow
+
+1. **Load & validate** -- Read the file, confirm shape and dtypes, report missing values.
+2. **Profile** -- Run `df.describe()`, check nulls, detect outliers (IQR or z-score).
+3. **Transform** -- Filter, group, pivot, or resample as needed.
+4. **Visualize** -- Generate charts; save to file with `plt.savefig()`.
+5. **Report** -- Summarize key findings in plain language.
+
+## Error Recovery
+
+| Problem | Action |
+|---------|--------|
+| File not found / wrong path | List directory contents, ask user to confirm filename |
+| Encoding error on read | Retry with `encoding='latin-1'` then `'cp1252'` |
+| Mixed dtypes in column | Use `pd.to_numeric(col, errors='coerce')` and report coerced rows |
+| Empty dataframe after filter | Warn user, show original value counts for filter column |
+| Chart rendering fails | Fall back to text-based summary table |
+
+## Example: end-to-end EDA
+
+```python
+import pandas as pd, matplotlib.pyplot as plt, seaborn as sns
+
+df = pd.read_csv('sales.csv', parse_dates=['date'])
+assert not df.empty, "Dataset is empty"
+
+# Profile
+print(df.describe())
+print(f"Missing values:\n{df.isnull().sum()}")
+
+# Visualize
+fig, axes = plt.subplots(1, 2, figsize=(12, 5))
+df.groupby('region')['revenue'].sum().plot.bar(ax=axes[0], title='Revenue by Region')
+sns.heatmap(df.select_dtypes('number').corr(), annot=True, ax=axes[1])
+plt.tight_layout()
+plt.savefig('eda_report.png', dpi=150)
+```