ag-ai-auditor

What problem does it solve?

AI-built Excel financial models contain unique, systematic errors that standard audits miss—like formulas stored as text, static value snapshots, broken cross-sheet links, and data type mismatches—that can lead to completely incorrect financial outputs if left undetected.

Core Features & Use Cases

• AI-Specific Error Detection: Identifies high-frequency errors introduced by LLMs using openpyxl, pywin32, or similar tools, including formula-as-text, static snapshots, uniform value fills, and broken sheet references.
• Automated Scanning: Runs a deterministic Python script to check all cells for the full catalogue of AI failure modes, with no manual data extraction required.
• Contextual Judgment Support: Includes a reference guide for human review of scanner results to filter false positives and validate high-impact findings.
• Use Case: Finance teams that use AI to build or modify financial models can use this skill to verify model accuracy before relying on outputs for forecasting, reporting, or decision-making.

Quick Start

Use the ag-ai-auditor skill to scan the attached Excel file 'ai-built-forecast.xlsx' for common AI-introduced errors and generate a full standardized audit report.