Industrial Inspection - Data Quality Audit

Overview

Automated row-by-row auditing of industrial quality inspection Excel files. Generates reports with only problematic cells highlighted in red, plus a summary statistics sheet.

When to Use

When the user needs to audit industrial quality inspection data, review Excel inspection reports, flag anomalous measurements, or produce red-marked inspection reports.

Workflow

1. Confirm Paths

Confirm with the user:

• Input directory: where the inspection Excel files are located
• Output directory: where reports will be saved (default: current workspace directory)

2. Run Inspection Script

Execute the bundled script:

python3 SKILL_DIR/scripts/audit.py <input_dir> <output_dir>

The script automatically:

• Detects and skips header/title rows in Chinese
• Audits every row against 4 core rules
• Highlights only problematic cells in red (not entire rows)
• Generates per-file inspection reports (no title row, no summary row, no extra columns)
• Generates a summary Excel with overview and per-file statistics

3. Present Results

After execution, show the user:

• Number of files reviewed, files with issues, total data rows, anomaly rows, overall pass rate
• For each anomaly: product code, test item, specific problem, and which columns are red-marked
• Paths to the generated report files

Core Audit Rules

Detailed rules are in references/rules.md. Summary:

| Rule | Description | Red-Marked Column | |------|-------------|-------------------| | Rule 1 - Range Exceedance | Measured value outside standard/reasonable range | Measured Value | | Rule 2 - Illegal Value | Negative or zero where not allowed | Measured Value | | Rule 3 - Deviation Anomaly | Absolute deviation rate exceeds 100% | Deviation Rate | | Rule 4 - Judgment Contradiction | Anomaly present but marked as "qualified" | Judgment Result |

Standard ranges are taken from the "Standard Requirement" column first; if empty, default reasonable physical ranges are used.

Column Structure

All inspection files are expected to have 10 columns in this order: 0. Product Code

1. Test Item
2. Test Standard
3. Measured Value
4. Standard Requirement (range)
5. Unit
6. Deviation Rate
7. Judgment Result
8. Test Date
9. Tester

Resources

scripts/audit.py

Main audit script. Runs independently with input and output directory arguments.

references/rules.md

Detailed audit rules with full standard ranges and anomaly judgment logic. Read when rule details are needed.