Acquire & validate
Download a complete date-and-product slice from the CFPB API. Check duplicates, dates, row counts, and source provenance.
From 7,953 CFPB money-transfer complaints to a transparent weekly signal review. Built to help an operations analyst ask better questions, faster.
Complaint counts show what people reported to CFPB. They are useful prompts for investigation, but they are not a customer incidence rate or proof of underlying harm.
Shares use only the complaints in this filtered dataset.
These flags compare a full week's issue count with the four prior full weeks. They are a screening queue, not a finding of root cause.
Download a complete date-and-product slice from the CFPB API. Check duplicates, dates, row counts, and source provenance.
Aggregate weekly issues with Python. Compare with a four-week baseline and surface only material jumps that pass documented thresholds.
A local retrieval layer answers questions from generated metric cards and cites the project files. MCP tools expose the same evidence to AI clients.
Inspect the Python, MCP tools, data, tests, and study guide →