data cleaning validation

The problem

Raw CSV and Excel exports from CRMs or surveys are often riddled with inconsistent casing, multiple date formats, and fragmented null values. Performing analysis on unvalidated data leads to silent errors and unreliable business insights.

What it does

• Identifies quality issues including mixed types, hidden null variants, and statistical outliers before touching the data.
• Generates a pre-execution cleaning plan for user approval to prevent accidental data loss.
• Standardizes text casing, whitespaces, ISO dates, and categorical values across the entire dataset.
• Executes rule-based validation for business logic like age ranges, email formats, and positive currency values.
• Produces a verifiable audit report with completeness scores and a line-by-line correction log.

Frameworks & tools

Compatible with CSV and Excel formats. Designed for use with Claude Code, Cursor, and other LLM-based coding environments.

Why this beats prompting it yourself

Standard prompts often "fix" data by guessing or silently deleting rows, which corrupts your audit trail. This skill enforces a diagnostic-first protocol that ensures every modification is logged, every ambiguity is flagged, and every resulting file is mathematically auditable.

Use cases

• Preparing messy CRM exports for migration into a production database.
• Cleaning multi-format survey results before generating executive reports.
• Validating financial datasets against business rules to flag anomalies.
• Standardizing legacy logs into ISO-compliant formats for time-series analysis.

Known limitations

Does not guess ambiguous dates or categories. Results rely heavily on provided or inferred business rules. Extremely high-volume files may require chunking depending on your environment context limits.