FedTrace
For any federal grant, contract, or lease: what was the stated purpose, what does the payment and audit record show, and what would a reasonable auditor conclude?
A methodology, not a watchdog. The record applies equally to any administration, any year, any program.
Notebooks
| # |
Notebook |
Run |
| 01 |
Reconstruct the financial record for cancelled federal awards — ceiling, obligated, outlays across public data sources |
 |
| 02 |
Full-scale fetch — pull raw API responses for all 13,440 contracts and 15,887 grants into resumable JSONL checkpoints |
 |
| 03 |
Assemble the three-number record — join IDV child amounts, compute ceiling / obligated / outlays for every matched award |
 |
| 04 |
Grant outlays investigation — probe three USASpending API endpoints; all confirmed dead ends for cancelled assistance awards |
 |
| 05 |
Grant outlays via bulk download — stream-parse Assistance_PrimeTransactions files, filter to matched FAINs, assemble grant three-number record |
 |
| 06 |
Combined accountability record — unified per-award schema joining contracts and grants; ceiling / obligated / outlays across all 21,873 cancelled awards |
 |
Setup
Add a GITHUB_TOKEN_FEDTRACE secret in Colab (key icon → left sidebar) before running any notebook that publishes results. Fine-grained PAT, resource owner fedtrace, Contents: Read + Write.