Status. Last reviewed 2026-08-19. Next review trigger: a new public dataset worth adding, a tool changing access terms, or a broken link below.
Our reporting rests on a simple principle: the accountability shouldn’t depend on trusting us. It should depend on the record being open. Almost everything we publish is built from public documents — financial disclosures, court filings, procurement records, corporate filings — that anyone can read. The work is mostly in knowing where they are and how to read them.
This page points you at the same materials we use, and at the open-source research infrastructure we publish so other reporters, organizers, and citizens can do the work too. If you find something we missed, that is the system working as intended. Tell us.
The databases we build on
The Capture Cascade Timeline
A searchable database of 4,500+ documented events from 1142 to the present, each with source citations — the spine that connects individual stories into patterns. When a piece of ours says “this is part of a longer arc,” this is the arc. Use it to trace a person, an institution, or a mechanism across time.
The ProPublica administration financial-disclosures database
A searchable record of 1,583 Trump-administration appointees and their financial disclosures — net worth, holdings, prior positions, agency — assembled by ProPublica from federal OGE-278 filings. This is the raw material behind some of our sharpest reporting: when a regulator’s old company turns up on the docket they now oversee, the disclosure is where you find it.
→ ProPublica’s Trump Town / appointee tracker
How to read it: the document type to learn is the OGE Form 278e — the public financial disclosure. The number that matters most is often not what someone holds but what they gave up to take the job: a forfeited earnout or divested position is a floor estimate of what the seat is worth. Two of our pieces show the method in action — The Regulator Who Wrote His Own Amicus Brief (a forfeited ~$18.8M earnout, then a rule mirroring the former employer’s Supreme Court brief) and The Recusal That Wasn’t (the gap between a sworn “recused” and the ethics agreement actually signed).
The Detention Pipeline early-warning map
A public heat map and county-by-county tracker of where new ICE detention facilities are likely headed next — built from commission activity, budget distress, real-estate traces, job postings, sheriff networks, and procurement signals, FIPS-coded so the data cross-references. If you want to know whether your county is being targeted, start here.
→ detention-pipeline.transparencycascade.org · and our explainer, Is Your County Being Targeted?
Pyrite — the knowledge base behind the reporting (open source)
Every investigation here runs on a structured, versioned knowledge base rather than a folder of documents: sourced entries with typed relationships, full-text and semantic search, and a QA pass that catches unsourced claims before they reach a draft. Notes live as plain markdown in git, so every change to the record is a reviewable commit.
Pyrite is the open-source system that does that. It was built here, by Transparency Cascade Press, because the reporting needed it — and it is agent-first by design: the AI research assistants that do the retrieval are treated as users of the knowledge base, not as bolt-on features. It has since grown past this newsroom into a general tool, and the code is there if you want your own.
→ pyrite.wiki · source at github.com/markramm/pyrite
Public records: where to start
- Federal contracts and spending — USAspending.gov is the master record of who the government pays and for what. Most of our procurement reporting (the detention-warehouse buildout, the defense-AI awards) starts with a PIID — a contract identifier — looked up here.
- Corporate filings — the SEC’s EDGAR holds 10-Ks, 8-Ks, and proxy statements; a company’s own filings often tell investors what its press releases won’t.
- Court filings — PACER (federal) and CourtListener (free, searchable) are where litigation, settlements, and sworn testimony live.
- FOIA — the federal FOIA.gov portal and MuckRock (which files, tracks, and publishes requests) are the tools for prying loose what isn’t posted. The record is often in the delay and the redaction as much as the documents.
Open-source investigation and digital safety
If you’re going to do this work, do it safely, and learn from the people who do it best.
- Bellingcat’s online investigation toolkit — the standard reference for open-source investigation methods.
- EFF Surveillance Self-Defense — operational security for researchers, organizers, and at-risk communities.
- Our own field guide — Make Yourself Expensive: shrinking your data footprint, for when the people you investigate can investigate you back.
A note on method
Read our editorial policies for how we verify, when we use anonymous sources (rarely), and how we handle requests for comment and corrections. The short version: primary documents first, multiple sources for contested claims, and we describe legal outcomes, not motives. If you do this work, hold yourself to the same standard — it is what makes it stick.
Maintained alongside our reporting. Found a dataset or tool we should add? contact@transparencycascade.org.