| name | officecli-data-dashboard |
| description | Use this skill when the user wants to create a data dashboard, analytics dashboard, KPI dashboard, or executive summary from CSV/tabular data in Excel format. Trigger on: 'dashboard', 'KPI report', 'analytics summary', 'data visualization', 'CSV to Excel dashboard', 'executive dashboard', 'metrics dashboard'. Output is always a single .xlsx file. |
Data Dashboard Skill
Create professional, formula-driven Excel dashboards from CSV or tabular data. The output is a single .xlsx file with a data sheet and a Dashboard sheet -- charts linked to live data, KPIs powered by formulas, and conditional formatting for visual insight.
BEFORE YOU START (CRITICAL)
If officecli is not installed:
This workflow expects OfficeCLI, but the skill must not install it automatically. Tell the user OfficeCLI is missing and present these choices before proceeding:
- Install forked OfficeCLI from
https://github.com/lidge-jun/OfficeCLI.
- Switch to the parent XLSX lightweight fallback path if the requested dashboard can tolerate lower fidelity.
- Stop or cancel.
Before switching to lightweight fallback, ask again and state which dashboard features may be lost, such as Office-native validation, chart/table fidelity, recalculation behavior, or CJK/rhwp-aware behavior. If the user chooses lightweight mode, save that preference to memory for future Office work.
Use upstream/vanilla iOfficeAI/OfficeCLI only when the task explicitly asks for vanilla upstream OfficeCLI behavior.
After user approval to install the fork:
bash "$(npm root -g)/cli-jaw/scripts/install-officecli.sh"
officecli --version
Use When
- User wants to create a dashboard from CSV data or tabular data
- User asks for KPI reports, analytics summaries, or metrics dashboards
- User wants to visualize data in Excel with charts, sparklines, and conditional formatting
- User mentions "CSV to Excel", "executive dashboard", or "data visualization"
What This Skill Produces
A single .xlsx file with:
| Component | Sheet | Description |
|---|
| Raw data | Sheet1 | Imported CSV with frozen headers, AutoFilter, column widths, conditional formatting |
| Dashboard | Dashboard | KPI cards (formula-driven), sparklines, charts (cell-range-linked), preset styling |
The Dashboard sheet is active on open. Formulas are stored as formulas and
fullCalcOnLoad should be set when available, but cached values are not proof
that Excel has recalculated. Before delivery, verify formula existence and run a
recalc/error check or record the exact spreadsheet app-open caveat.
Core Concepts
Formula-Driven KPIs
Every KPI value on the Dashboard is a formula referencing the data sheet. Never hardcode calculated values. When the underlying data changes, KPIs update automatically.
Cell Range References for Charts
Every chart series references data sheet cells directly (series1.values="Sheet1!B2:B13"). Charts stay in sync with data. Never use inline data unless aggregation is impossible in Excel formulas.
Chart Presets
Use preset=dashboard on charts for datasets with 10+ rows. For datasets with fewer than 10 rows, use preset=minimal. See the complexity table in A.3 of creating.md for the authoritative mapping -- when any other text in this skill conflicts with that table, the table wins. Presets are DeferredAddKeys -- they work on add only, NOT on set. A single preset replaces 5-8 manual styling properties with one consistent look.
Data-Size-Aware Complexity
The number of KPIs, charts, sparklines, and CF rules scales with the input data size. A 5-row dataset gets 1 chart and no sparklines. A 200-row dataset gets 3-5 KPIs, 2-3 charts, sparklines, and multiple CF rules.
Workflow Overview
Phase 1: Analyze the Input Data
Count rows and columns. Identify column types (date, numeric, categorical). Determine the primary dimension (X-axis). Look up the data-size-to-complexity table.
Phase 2: Plan Before Building
Decide how many KPIs, which chart types, which CF rules, and chart layout positions. Write out the plan before executing any commands.
Phase 3: Build the Workbook
Follow the 11-step workflow: create + import, column widths, Dashboard sheet, KPIs, sparklines, charts, conditional formatting, tab colors, polish, raw-set, validate.
Phase 4: QA
Run the QA checklist. Fix issues. Re-validate.
Phase 5: Deliver
Deliver the .xlsx file. Tell the user the Dashboard sheet opens first and formulas recalculate automatically.
Full Guide
Read creating.md and follow it step by step. It contains the complete workflow, decision tables, command templates, a full runnable example, and the QA checklist.
Quick Reference: Key Warnings
| Warning | Detail |
|---|
| Batch invocation | cat <<'EOF' | officecli batch dashboard.xlsx — file argument is required. officecli batch alone fails with "Required argument missing". |
| Batch JSON values | ALL values must be strings: "true" not true, "24" not 24 |
| Chart preset | Add-only. preset=dashboard for 10+ rows, preset=minimal for < 10 rows |
| Scatter charts | Use series1.xValues NOT series1.categories (causes validation error) |
| Reference lines | Format is value:color:label:dash (color BEFORE label) |
| Cell range refs | Always series1.values="Sheet1!B2:B13", never inline data |
| raw-set ordering | activeTab raw-set must be the LAST command. Use set / --prop calc.fullCalcOnLoad=true for calcPr (not raw-set). |
| formulacf | Do NOT use font.bold. Use fill + font.color only |
| Column widths | import --header does NOT auto-size. Set widths manually on ALL sheets including Dashboard |
| Dashboard ### | KPI cells at 24pt bold WILL show ### if Dashboard columns are not set to width=22. See Step 4b |
| pie chart | chartType=pie has a known rendering issue in LibreOffice — chart renders blank (only legend visible). Use chartType=doughnut or chartType=column instead for category breakdowns. |
| Dashboard KPI placement | KPI labels and values MUST be on the Dashboard sheet itself, not on Summary/Data sheets. Dashboard must show: KPI area (rows 1-2) + charts (rows 5+). See Step 3 CRITICAL box in creating.md. |
| Helper columns | After using helper columns on Dashboard for chart data sources, hide them with officecli set file.xlsx '/Dashboard/col[N]' --prop hidden=true. Visible helper columns are unprofessional clutter. |
| Chart data source (CRITICAL) | Chart series MUST reference visible cells on the Summary sheet or Data sheet — NEVER reference hidden columns or helper columns. LibreOffice does not recalculate hidden columns on render; charts referencing hidden columns will display blank/empty data. Correct pattern: aggregate all chart data into visible rows on the Summary sheet, then reference those cells as chart data sources. |
| Charts within print area | After setting the print area, verify all charts are positioned inside it. Chart x + width must not exceed the print area column boundary. Recommended: chart width ≤ 90% of print area width. |
| Summary sheet col widths | Summary sheet columns are NOT auto-sized. Always set widths after writing formulas: text/label cols 15-20, numeric cols 12-15. See Step 2 in creating.md. |
| Percentage numFmt on Summary | AVERAGEIFS/ratio formula results on Summary sheet display as raw floats (0.083) unless numFmt="0.0%" is set at the same time as the formula. Never skip numFmt on percentage cells. |
| Header row fill (MANDATORY) | Every sheet with column headers MUST have a dark fill on row 1: officecli set dashboard.xlsx '/Sheet1/A1:F1' --prop fill=1F3864 --prop font.color=FFFFFF --prop font.bold=true. Omitting header fill is a Quality Bar violation (Q2 fail). See Step 2b in creating.md. |
| Conditional formatting (MANDATORY for 10+ rows) | At least 1 CF rule (colorscale or databar) MUST be applied to a numeric column in the Data sheet for datasets with 10+ rows. Verify presence with officecli query dashboard.xlsx 'conditionalformatting'. Zero CF rules is a Q4 quality failure. |
| KPI card background fill (Quality Bar) | Dashboard KPI card area (rows 1-2) SHOULD have a light background fill, e.g., fill=F0F4FF. Omitting this is a P2 deduction. See Step 4b in creating.md. |
Adjustments After Creation
When the user requests changes after the dashboard is built:
| Request | Command |
|---|
| Swap two sheets | officecli swap dashboard.xlsx '/Dashboard' '/Data' |
| Move a sheet after another | officecli move dashboard.xlsx '/Summary' --after '/Dashboard' |
| Edit a cell value | officecli set dashboard.xlsx '/Dashboard/A1' --prop value="..." |
| Find & replace text | officecli set dashboard.xlsx / --prop find=OldText --prop replace=NewText |
| Re-point chart data | Not supported — chart series are fixed at creation (data/dataRange are add-only). Delete the chart and re-add it with the new --prop dataRange=Sheet!A1:D10. |
References
- creating.md -- Complete dashboard creation guide (the main skill file)
- xlsx SKILL.md -- General xlsx reading, editing, and QA reference
