Research Datasets

Every curated Profoundd archive, published as a downloadable dataset — one JSONL file per collection, regenerated automatically as the archives grow. Free for research under CC-BY 4.0 (attribute "Profoundd, profoundd.com").

Poll manifest.json to detect updates cheaply — it lists every collection's SHA-256 checksum, document count, and last-updated timestamp. Transcripts of video/audio are verbatim ASR captions, not certified transcripts; documents keep their original source URLs for verification.

Dataset	Documents	Size	Updated	Download
FBI Vault	8,895	438.7 MB	2026-09-02	JSONL
Southern Poverty Law Center (Wayback)	7,830	6.4 MB	2026-09-02	JSONL
Pfizer Documents (PHMPT/FDA)	2,373	225.2 MB	2026-09-02	JSONL
CDC ACIP — Vaccine Advisory Committee	449	9.5 MB	2026-09-02	JSONL
NIH Pandemic-Era Grants	9,873	78.8 MB	2026-09-02	JSONL
Survival, Water, Medical Field Manuals	1,435	98.5 MB	2026-09-02	JSONL
Rudolf Steiner Archive	342	65.4 MB	2026-09-02	JSONL
Hesperian Health Guides (Where There Is No Doctor)	19	3.2 MB	2026-09-02	JSONL
Robots — Open-Source 3D-Printable Designs	27	22.5 KB	2026-09-02	JSONL
Smart Cities & Surveillance — WEF, Contracts, FOIA	151	164.2 KB	2026-09-02	JSONL
Fauci Files — DNI Gabbard Release + Rand Paul Diaries (2026)	50	6.8 MB	2026-09-02	JSONL
Mail Tribune (Medford, OR — Wayback)	144,941	761.7 MB	2026-09-02	JSONL
Ashland Daily Tidings (Ashland, OR — Wayback)	32,377	190.7 MB	2026-09-02	JSONL
Charlie Kirk / Tyler Robinson Case — Court Transcripts & Filings	270	21.4 MB	2026-09-02	JSONL
Medical Talks — Integrative & Longevity Medicine	771	13.1 MB	2026-09-02	JSONL
Robert Malone — Collected Works	233	4.2 MB	2026-09-02	JSONL
Peter McCullough — Collected Works	169	725.5 KB	2026-09-02	JSONL
Pierre Kory — Collected Works	317	6.1 MB	2026-09-02	JSONL
Oregon LUBA — Land Use Board of Appeals Oral Arguments	36	1.0 MB	2026-09-02	JSONL
Iran "Lego" Propaganda — Under Analysis	12	26.0 KB	2026-09-02	JSONL

How to use

Each file is newline-delimited JSON — one document per line with title, content (full text), source_url, doc_date, doc_kind, and tags. Load with pandas.read_json(url, lines=True), stream with jq, or ingest into any search/analysis tool. Collections marked commentary/news/critique in doc_kind are labeled third-party material indexed for comparison — not the primary record.