SomaliScan is a public-data project that compiled US government spending and related public records into an open archive. Originally developed as a transparency platform with a database, search interface, and AI query tools, the project has concluded its live service and released its final data snapshot for public use. The archive contains 60 tables and approximately 696 million rows, provided as 37 GB of Apache Parquet files under the CC0 public-domain dedication.

Its datasets cover federal, state, and local spending, campaign finance, lobbying, and federal employment. Highlights include state vendor payments across all 50 states, federal campaign contributions linked to spending recipients, Paycheck Protection Program loans cross-referenced with federal contractors, and CMS Open Payments records. Users can browse the archive on Hugging Face, access its source code on GitHub, consult a query cookbook, or run queries directly against the hosted Parquet files using DuckDB, without downloading the full collection or creating an account. SomaliScan also provides citation details for the dataset.