FCHDS Archive

Holds about 91,000 documents and 144 data tables from the FCHDS investigation.

Ask a question. Type a question in plain words. The page finds the matching passages, writes a short answer, and lists the exact quoted lines it relied on, each checked word for word against its source. Follow-up questions remember the topic. Written answers are capped at 130 a day.

Search. Exact words, names, case numbers, or a phrase in quotes. Filter by kind of document (court and FOIA files, map cards, research files and more). Each result shows the matching passage. No daily limit.

Read a document. Open any result to read the whole document part by part, with a find-in-page box.

Browse the tables. The Tables tab holds the structured data: visitor logs, FOIA emails, court dockets, Gates grants, the Fauci diary, NIH panel records, FDA production data, World Health Summit speakers and sessions. Peek at a table's columns, search its text, or run a read-only SQL query (up to 200 rows; 400 queries a day; the 8-million-row visitor log needs a WHERE clause).

Use it from claude.ai. Add https://fchds-data.lelandlehrman.workers.dev/mcp as a custom connector in claude.ai. Claude can then search, read documents and query the tables from any chat. Tools: search_fchds, read_document, list_tables, describe_table, sql_select, search_table, list_source_types.

Use it from a program. JSON endpoints: /api/search?q=, /api/doc?id=, /api/ask?q=, /api/tables, /api/describe?t=, /api/sql?q=, /api/tsearch?table=&q=, /api/types.

Your privacy. Recent searches are stored in your own browser only.

What to keep in mind. This is a read-only copy of the laptop archive, refreshed by hand, so the newest documents appear after the next export. Written answers come from a small built-in AI model that only sees the passages it found: open the sources before relying on anything.