Project overview
I developed a skill suite for agents such as Claude and Cursor focused on strategic credit card expense analysis. The flow separates extraction integrity from analytical quality: first it consolidates invoices and statements from multiple sources into a traceable, reconciled CSV; then it categorizes transactions, calculates metrics, and generates a self-contained HTML dashboard with recommendations prioritized by impact and effort.

Challenge
Turn PDFs, OFX files, CSVs, and images from different banks into a reliable decision-making dataset while avoiding extraction errors, duplicates, total mismatches, and shallow analysis that does not clarify what to cut first.
Solution
Two complementary skills: consolidar-faturas standardizes, reconciles, and logs transactions; analise-estrategica-despesas applies categorization rules, calculates metrics, and delivers a phased plan for immediate cuts, containment, optimization, and review.
Tech Stack
- Claude
- AI Agents
- Python
- Financial Analysis
- Dashboard
Technical scope
- PDF, OFX, CSV, and image consolidation
- Reconciliation, deduplication, and audit trail
- Categorization and impact-versus-effort prioritization
- HTML dashboard with a strategic savings plan
