csv-parser_skill

Overview

This skill parses and analyzes CSV files to produce validated, actionable summaries. It detects delimiters and encoding, infers column types, reports missing or malformed data, and computes basic statistics. It is designed to handle large or imperfect CSVs gracefully and surface issues for downstream processing.

How this skill works

The skill reads a CSV file, attempting multiple delimiter and encoding heuristics to find the best match. It inspects each column to infer data types, validate values against common constraints, and collect counts of missing or malformed entries. Numeric columns receive summary statistics (mean, median, min, max, standard deviation) while categorical columns receive frequency summaries. The tool returns a structured analysis report including row/column counts, validation flags, and suggested fixes.

When to use it

• You need a quick integrity check before importing CSV data into databases or ML pipelines.
• You want automatic detection of delimiter and encoding for unknown or inconsistent CSV exports.
• You need summary statistics and data type inference to inform schema design.
• You have CSVs with possible malformed rows, mixed types, or missing values.
• You want to generate a validation report to share with data providers.

Best practices

• Run the parser on a representative sample first to confirm delimiter and encoding assumptions.
• Provide a small schema or expected types when possible to improve validation accuracy.
• Review the missing-value and malformed-data report before bulk processing or automated ingestion.
• Use the frequency summaries to decide on categorical encoding or trimming rare categories.
• Re-run validation after applying fixes to confirm issues were resolved.

Example use cases

• Preflight check for CSV exports from third-party systems to detect encoding or delimiter problems.
• Automated data validation step in ETL pipelines to reject files with too many malformed rows.
• Exploratory data analysis to quickly learn column types, missingness, and summary statistics.
• Data quality reports for suppliers showing validation failures and suggested corrections.
• Preparing training data for ML by identifying numeric vs categorical columns and outliers.