pretty-but-wrong-detector

Audit polished decks and workbooks for unsupported claims and formula errors.

Updated Jul 16, 2026
One-click install
npx skills add https://github.com/Cloud-Byte-Consulting/plugins --skill pretty-but-wrong-detector-cloud-byte-consulting
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pretty-but-wrong-detector
Source: https://github.com/Cloud-Byte-Consulting/plugins/tree/main/prompt-workflows/skills/pretty-but-wrong-detector
Command: npx skills add https://github.com/Cloud-Byte-Consulting/plugins --skill pretty-but-wrong-detector-cloud-byte-consulting

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI-generated or heavily revised presentations and spreadsheets often look finished while hiding unsupported claims, untraceable numbers, and broken formulas that busy reviewers may forward without checking. ## Core Features & Use Cases - Skeptical Claim Audit: Flags claims without source attribution, numbers without dates or sources, and charts with untraceable underlying data. - Formula Integrity Checks: Detects inconsistent formulas across parallel rows, hardcoded outputs where formulas are expected, and fixed cell references that should roll forward. - Ranked Issue Report: Produces a written list of every issue ranked as must-fix, should-fix, or polish, ending with a readiness verdict of not ready, ready with limitations, or ready to share. - Use Case: Before forwarding an AI-drafted quarterly workbook to leadership, run this audit to catch stale date ranges, assumptions presented as facts, and broken narrative logic. ## Quick Start Use the pretty-but-wrong-detector skill to audit this attached workbook for unsupported claims and formula errors before I share it.

Frequently Asked Questions about pretty-but-wrong-detector

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I check an AI-generated presentation for errors before sharing?

Run a skeptical review pass that flags claims without sources, numbers without dates, and charts with untraceable data. This skill enumerates every issue, ranks severity, and gives a readiness verdict without changing the file.

How to audit Excel formulas for hardcoded values and broken references?

Review each sheet for formulas inconsistent across parallel rows, fixed cell references that should roll forward, and hardcoded outputs where formulas are expected. Also test whether outputs change when assumptions change.

Does this skill fix the issues it finds in my deck or workbook?

No, it only enumerates and ranks issues as must-fix, should-fix, or polish. It deliberately does not modify anything, leaving all corrections to human judgment.

When should I run a document audit before sharing?

Run it when a deck or workbook looks finished, was generated or heavily revised by AI, and might be trusted by a busy recipient because it looks clean. It is designed as a final gate before forwarding.

What readiness verdicts does the audit produce?

The audit ends with one of three verdicts: not ready, ready with limitations, or ready to share. It will not mark a document ready if any must-fix issue remains.