| name | analyze-my-data |
| description | Runs a plain-language business analysis of uploaded data, producing key findings, data quality notes, charts, and suggested next questions for non-technical users. |
Analyze my data
Use this skill when the user wants Orion to inspect a business data file without needing to choose an analysis method first.
Workflow
- Identify the referenced file or ask which uploaded file to analyze.
- Load the data in Python with conservative parsing. For spreadsheets, inspect sheet names first and pick the most relevant sheet unless the user specified one.
- Summarize the dataset in plain language: what each likely column means, row count, date range, important categories, and obvious data quality issues.
- Run focused exploratory checks:
- totals and trends over time;
- top and bottom categories, products, customers, stores, or channels;
- missing values, duplicates, impossible values, and outliers;
- relationships that matter for the apparent business context.
- Create notebook outputs that a non-technical user can read: concise markdown, clear charts, and interactive tables where helpful.
- If the analysis produces a useful report, load
create-app and mark the summary, charts, and tables for App View.
Tone
Write for a business owner or finance/ops teammate. Avoid statistical jargon unless it is necessary, and explain it in one sentence when used.
Completion
Finish with:
- 3-7 key findings;
- data quality caveats;
- recommended next actions;
- a short list of follow-up questions the user can ask Orion.