data-report · Data Visualization Reports
data-report turns CSV, Excel, or JSON data into a complete visualization report page: scripts compute KPIs and pick 2–4 charts by data shape, assembling a self-contained HTML page with KPI cards, charts, insights, and a data table — viewable offline.
KPIs and charts are computed deterministically from real data, never hand-written. Adding 3–5 interpretive insights (trends, anomalies, comparisons, recommendations) is recommended; the page can also be rendered as a shareable long image.
Example invocation: "Turn this sales data into a visual report."
Full brief
Positioning
data-report is a report-page generator for structured data: CSV, Excel, or JSON in, a full visualization page out (KPI cards + charts + insights + data table). Parsing, aggregation, charting, and HTML assembly are all done deterministically by scripts — no single-chart or infographic output.
Core capabilities
- One command to a full page:
scripts/report.pycomputes KPIs (numeric-column summaries) and picks 2–4 charts by data shape (time series → line, categories → bar, share → pie) into one HTML page. - Overview first:
analyzereturns row/column counts, column types, numeric summaries, and Top 5 categories — the factual basis for insight writing. - Self-contained & offline: charts embedded as base64; a single HTML file opens anywhere with no garbled CJK text.
- Insight pass: 3–5 interpretive notes (trends, anomalies, comparisons, recommendations) based on the real overview — interpretation only, never invented numbers.
- Shareable long image: optionally render the page to a 1080px-wide image for social sharing.
- Profile theming: with a profile, brand colors from
style.mdare applied. - Determinism: real data through the script, always; KPIs and charts are never hand-written. Without numeric columns, the data table still renders.
Workflow
- Inspect:
report.py analyze <data file>— figure out what the data says - Generate:
report.py report(optional title and KPI columns) - Add insights (optional, recommended): 3–5 interpretive notes, no invented numbers
- Render a long image for sharing (optional)
Inputs & outputs
| Input | Notes |
|---|---|
| Data file | .csv / .json / .xlsx (Excel needs openpyxl; the script warns if missing) |
| Report title | Optional |
| KPI columns | Optional — auto-picked from numeric columns otherwise |
Output: self-contained HTML report (to outputs/), optional PNG long image.
Boundaries with adjacent skills
- chart-visualization: calls the AntV remote API for a single static chart image URL — for one standard chart as a direct link.
- infographic: local JS rendering for structured infographics / animated GIF charts.
- data-report: a full-page report from a data file.
Fit
- Business weekly/monthly report pages
- Marketing and ops post-mortems on campaigns and KPIs
- Turning raw data into a shareable visual report
- Analysts quickly validating data shape and capturing insights
Before you start
- No API key; everything runs locally.
- Excel input needs openpyxl (the script warns if missing).
- Accuracy depends on truthful input; without numeric columns the data table still ships.
data-report is part of the Aiglade Skill library. Invoke it from the Aiglade chat box in plain language.