Dashboard Caption Audit
Audits an open Tableau workbook for missing or default titles and captions across dashboards, worksheets, dashboard text objects, and phone / tablet device-layout variants.
Example: Ask for a caption audit on a workbook where half your worksheets still say "Sheet 1" and three dashboards ship with the placeholder <Sheet Name> text, and you get back a prioritized eight-rule report with exact locations and manual-fix steps.
What it does
Scans every dashboard, worksheet, and text object in an open Tableau workbook for title and caption hygiene against an eight-rule completeness catalog. Catches missing titles, missing captions, empty text objects, default names in English / French / Spanish / German, device-layout inheritance duplicates, and placeholder-title text like <Sheet Name> that Tableau leaves in when you disable auto-title. Read-only. Every finding is a manual Desktop-fix instruction; you apply the fix. Two-tier delivery: compact chat summary plus opt-in file save.
When to use it
- "I have a workbook that ships with 'Sheet 1' worksheet titles and I want to catch them before publish"
- "I want to run a monthly QA sweep across a governance library for caption completeness"
- "I'm auditing device-layout variants (phone / tablet) and want to know which ones inherit stale defaults from desktop"
- "I need a prioritized list of caption gaps for a workbook I'm about to hand off"
- "I want to verify a workbook has no placeholder title text before it goes to a stakeholder"
When NOT to use it
- Not for spellcheck or typography (use
desktop-spellcheckfor text hygiene) - Not for tooltip content audit (out of scope)
- Not for caption position or sizing sanity (out of scope)
- Not for format-string drift on caption fonts (use
format-drift-scanner) - Not for dashboard assembly hygiene like zone alignment or filter-action wiring (use
dashboard-preflight) - Not for calc-family metadata audit (use
calc-linter)
How to trigger it
- "caption audit"
- "check my titles and captions"
- "audit dashboard captions"
- "dashboard-caption-audit"
- "find missing titles in this workbook"
What you'll need
- Tableau Studio with the target workbook open
- Every relevant dashboard and worksheet accessible via
get-dashboard-xmlandget-worksheet-xml - A Downloads path or user-specified folder for the optional file output
Example walkthrough
Input: "caption audit" on a workbook with 3 dashboards, 12 worksheets, phone layouts on 2 dashboards
Tier A chat output:
Caption Audit Summary
Workbook: SalesReview_Q3.twbx
Dashboards: 3 | Worksheets: 12 | Device layouts: 2 phone
Findings: 7 Warning, 3 Info
TL;DR (top 3 Warnings):
1. Dashboard "Executive Overview" phone layout inherits default "Sheet 1" title
from desktop but desktop title was manually set to "Q3 Sales Rollup"
(DC-DEVICE-07)
2. Worksheet "Untitled" retains default name across 4 dashboard placements
(DC-DEFAULT-NAME-06)
3. Dashboard "Regional Detail" title reads "<Sheet Name>" placeholder text
(DC-TITLE-08)
Full rule breakdown:
- DC-TITLE-01 Dashboard title missing: 0
- DC-CAPTION-02 Dashboard caption missing: 2 (Info)
- DC-WS-TITLE-03 Worksheet title missing: 1 (Warning)
- DC-WS-CAPTION-04 Worksheet caption missing: 1 (Info)
- DC-TEXT-EMPTY-05 Text object empty: 0
- DC-DEFAULT-NAME-06 Default name: 4 (Warning)
- DC-DEVICE-07 Device-layout inheritance duplicate: 1 (Warning)
- DC-TITLE-08 Placeholder title text: 1 (Warning)
Save full report to file?
Tier B (opt-in file output, after path confirmation): a .md document with per-dashboard sections, per-worksheet findings, exact Desktop-fix instructions per rule, and a handoff YAML block for downstream skills.
What you get back
- Compact chat summary (Tier A): totals, top 3 Warnings sorted deterministically, per-rule breakdown
- Optional full report (Tier B): every finding with exact location and Desktop-fix instruction
- Handoff schema
caption_audit_report/1for downstream skills (dashboard-preflight, desktop-spellcheck, format-drift-scanner, viz-planner, workbook-auditor) - Deterministic ordering: same input produces byte-identical top-3 across runs (HC#14 + HC#15)
Limitations
- One workbook per run. Bulk-workbook runs are refused.
- Locale coverage for default-name detection is English, French, Spanish, and German. Other languages will fall through to non-detection.
- Tooltip content is not audited (use
caption-and-tooltip-advisorfor tooltip rewrites). - Caption position and sizing are not audited (out of scope).
- Spellcheck of caption text is out of scope (defer to
desktop-spellcheck). - Font-family or font-size drift in captions is not detected here (defer to
format-drift-scanner).
Troubleshooting
- "get-worksheet-xml failed on dashboard-bound sheet" then hits a known reliability gap on dashboard-bound worksheets. Skill emits
"unresolved: xml-read-failed"per HC#7 and continues. That one sheet's caption findings will be missing; the rest of the audit is intact. - "Placeholder title text not caught" then the placeholder pattern is outside DC-TITLE-08's regex.
- "Default name not caught in language X" then the locale isn't in the four-language regex list. English, French, Spanish, German covered.
- "Save aborted, no path confirmed" then HC#12 blocked a silent file write. Confirm a Downloads path when prompted.
- "Same top-3 not identical across runs" should not happen; V4 gate enforces deterministic ordering. If it does, file a bug with both run outputs.
Support
File issues in the Tableau Agent Skills repo.