Example: Fully Clean Report
This is the literal, complete output shape for a workbook with zero findings. Copy this structure exactly — substitute the real workbook name, date, allow-list count, and sheet count. Do not omit any section, do not add narrative, do not mention terminology consistency or any other out-of-scope observation.
# Spellcheck Report: 01_clean
Scanned: 2026-08-26 14:32
Categories included: Worksheet titles/captions, dashboard text/titles, field aliases, calc names, parameter labels, tooltips, annotations
Allow-list: skill-expertise/allow-list.txt, 110 terms loaded
Dictionary: en-US
Findings summary:
- Critical: 0
- High: 0
- Medium: 0
- Low: 0
- Total: 0
Sheets scanned: 26 of 26
### Critical
None
### High
None
### Medium
None
### Low
None
## Not Scanned
None. All sheets scanned successfully.
## Allow-list suggestions
None. No repeated unknown tokens.
***
Report generated by desktop-spellcheck. All findings are advisory. Fix in Tableau Studio manually. Re-run after edits to confirm.
Same example, chat rendering (no markdown syntax)
The chat panel does not render markdown. When posting in chat instead of writing to a file, use this plain-text form instead — same content and section order, no #, ##, **, or ---:
SPELLCHECK REPORT: 01_clean
Scanned: 2026-08-26 14:32
Categories included: Worksheet titles/captions, dashboard text/titles, field aliases, calc names, parameter labels, tooltips, annotations
Allow-list: skill-expertise/allow-list.txt, 110 terms loaded
Dictionary: en-US
Findings summary:
- Critical: 0
- High: 0
- Medium: 0
- Low: 0
- Total: 0
Sheets scanned: 26 of 26
CRITICAL
None
HIGH
None
MEDIUM
None
LOW
None
NOT SCANNED
None. All sheets scanned successfully.
ALLOW-LIST SUGGESTIONS
None. No repeated unknown tokens.
Report generated by desktop-spellcheck. All findings are advisory. Fix in Tableau Studio manually. Re-run after edits to confirm.
What this example demonstrates
- The Findings summary block with all four counts (even at zero) is never dropped, regardless of how "obviously clean" the workbook is.
- Every severity tier gets its own heading with "None" stated explicitly — headings are never skipped.
- "Not Scanned" and "Allow-list suggestions" always appear, even when empty.
- No terminology-consistency asides, no "worth your attention" notes, no offers to widen scope, no editorializing. If something like that comes to mind while scanning, it does not belong in this report — leave it out entirely.
- No tool names, MCP call sequencing, or internal mechanics anywhere in the output.