Guides
Skill Interaction Matrix
Skills are designed to chain together. The agent automatically transitions between skills based on your prompt — you don't need to name skills explicitly.
Common Chains
| Chain | Skills | When to Use |
|---|---|---|
| Load → Profile | magic-data-loading → magic-data-profiling | Assess quality of a new dataset |
| Load → Profile → Report | magic-data-loading → magic-data-profiling → magic-report-generation | Full quality audit with documentation |
| Load → Clean → Validate | magic-data-loading → magic-data-cleaning → magic-data-validation | Fix and verify data quality |
| Load → Explore → Visualize | magic-data-loading → magic-data-exploration → magic-data-visualization | Understand and visualize patterns |
| Load → Transform → Export | magic-data-loading → magic-data-transformation → magic-data-loading | Reshape and export data |
Skill Dependency Graph
graph LR
load[magic-data-loading] --> profile[magic-data-profiling]
profile --> clean[magic-data-cleaning]
clean --> validate[magic-data-validation]
validate --> transform[magic-data-transformation]
transform --> viz[magic-data-visualization]
transform --> report[magic-report-generation]
load -.-> explore[magic-data-exploration]
explore -.-> stats[magic-statistical-analysis]
explore -.-> viz
stats -.-> report
synth[magic-data-synthesis] -.-> validate
style load fill:#dbeafe,stroke:#3b82f6
style profile fill:#d1fae5,stroke:#10b981
style clean fill:#d1fae5,stroke:#10b981
style validate fill:#d1fae5,stroke:#10b981
style transform fill:#fef3c7,stroke:#f59e0b
style viz fill:#fce7f3,stroke:#ec4899
style report fill:#fce7f3,stroke:#ec4899
style explore fill:#e0e7ff,stroke:#6366f1
style stats fill:#e0e7ff,stroke:#6366f1
style synth fill:#f3e8ff,stroke:#a855f7Solid arrows = common chains (agent transitions automatically). Dashed arrows = alternative paths.
Data Flow Between Skills
Each skill consumes and produces specific data formats:
| Skill | Consumes | Produces |
|---|---|---|
magic-data-loading | Raw files (CSV, Excel, Parquet, JSON) | pandas DataFrame + checkpoint |
magic-data-profiling | DataFrame | Quality score, issue list, statistics |
magic-data-cleaning | DataFrame + issue list | Cleaned DataFrame + change log |
magic-data-validation | DataFrame | Validation report (pass/fail per rule) |
magic-data-transformation | DataFrame | Transformed DataFrame + checkpoint |
magic-data-exploration | DataFrame | Summary statistics, correlations, patterns |
magic-data-visualization | DataFrame | PNG charts + generation script |
magic-report-generation | Profiling/analysis results | Markdown report + validation |
magic-statistical-analysis | DataFrame | Test results, models, interpretations |
magic-data-synthesis | Schema or existing data | Synthetic DataFrame |
Was this page helpful?
Edit on GitHub
Last updated on