A dashboard for observing U.S. State Deptartment post counts

Detects content changes in 383 federal government documents using Wayback Machine captures. Updated weekly.

Archive and report government documents to the GPO.

Multi-archiver forensic audit of the 2024 End Of Term web crawl. Deduplicating billions of government records.

Two evaluations of automated metadata-assertion pipelines: (1) Interpolation — an LLM's ability to access and reason over tables and charts inside archival PDFs; (2) ANT Catalog — the ability to create document metadata, especially government classification. Both benchmark open and frontier models on Internet Archive / government collections.