visa-validator

Purpose

Scan a folder of visa documents, validate against country checklist, produce gap analysis and approval probability score.

Steps

1. List all files in the target folder
2. Read and extract text from each document via OCR
3. Load checklist from /checklists/{country}.json
4. PII Scrubbing Layer eg. Strip raw passport numbers before sending to LLM — send structure only
5. Compare docs vs checklist
6. Generate structured report with Gaps + Probability Score + Key Insights
7. Save report to folder as visa-review-report-{Current Date and Time}.md

Process

1. Input Collection: Ask user for the folder path containing visa documents and the target country
2. File Discovery: List all files in the specified folder
3. Document Processing: For each document:
   • Extract text using OCR (try multiple methods: macOS Vision, EasyOCR, Tesseract)
   • Apply PII scrubbing to remove sensitive data like passport numbers, SSNs, etc.
   • Structure the extracted data for comparison
4. Checklist Loading: Load the appropriate country checklist from /checklists/{country}.json
5. Validation & Gap Analysis: Compare document contents against checklist requirements
6. Scoring: Calculate approval probability based on completeness and accuracy
7. Report Generation: Create a detailed markdown report with:
   • Identified gaps
   • Approval probability score (0-100%)
   • Key insights and recommendations
8. Output: Save report as visa-review-report-{timestamp}.md in the same folder

Inputs

• Folder path containing visa documents
• Target country (for checklist selection)

Outputs

• Markdown report file: visa-review-report-{timestamp}.md

Dependencies

• OCR tools (macOS Vision, EasyOCR, Tesseract)
• JSON checklist files in /checklists/ directory
• Basic text processing utilities

Guardrails

• PII scrubbing must be applied before any LLM processing
• Handle missing checklists gracefully
• Support common document formats (PDF, JPG, PNG, etc.)
• Provide clear error messages for unsupported formats