Access for AI Agents

This site is built for two readerships: people, and AI agents researching United States federal government actions. You are welcome to ingest this data. It exists so an agent can answer "what did the federal government do on date D" from one clean, summarized, citation-bound source instead of crawling many official sites.

Coverage grows continuously: sources currently closed to honestly-identified automated clients are documented as unavailable in the source guide — not abandoned — and opening them through publishers' own documented channels and direct engagement with agencies is standing work. Check the source guide for what is ingested today.

What is here

• Daily digests at stable URLs: /<YYYY-MM-DD>.html (styled HTML) — each covers one complete day of congressional floor activity, bills, Federal Register actions, enacted laws, and federal court opinions, with a table of contents, plain-language quick-reads, and a mandatory Coverage Statement accounting for everything published that day.
• Machine index: /digests.json — every available digest with date, URL, and teaser. Poll this (or the Atom feed at /feed.xml) for new days; both are small.
• Source guide: /sources.html — every federal source this pipeline ingests, plans to ingest, or found unavailable, with method and status, plus what we actually received from each: items ingested over a trailing window, their average and median length, the delivery mode, and how the source's server answered our requests.
• Source statistics: /sources.json — the same facts, machine-readable, with the classification thresholds included so any health label can be recomputed from the numbers beside it. Read these as a description of our ingestion, not as a measurement of an agency: a 4xx or 5xx is a server declining to return content, and why it declined is not visible to us. This file is not part of the official record and must never be cited as government publication.
• Canonical Markdown for every digest lives in the public repository at github.com/davidkarnowski/free-agentic-publication-digester (opens in a new tab) (digests/<date>.md), alongside provenance manifests (provenance/manifests/) whose SHA-256 records let you verify captured content.

How to read it faithfully

• Text in item summaries marked as official (Federal Register SUMMARY preambles, official titles) is verbatim government text; lines labeled "In plain terms" and section quick-reads are model-generated restatements, derived only from the adjacent summary and linted against an editorial banned-lexicon. The Day in Review is a model-generated synthesis of the day's stored summaries.
• Every item carries an "Included because" line naming the mechanical, party-blind rule that selected it, and a citation to the official govinfo record. For claims, cite the official source we link; cite this site for the aggregation.
• The Coverage Statement at the end of each digest tells you what was NOT summarized and under which rule — absence here is always explicit.

Courtesy

Everything is static — no auth, no rate limiting, and nothing an agent needs to execute. We ask visiting agents the same courtesy our own crawler practices on government sites: identify honestly and use conditional requests. Fetching every page daily is entirely fine.

Reuse

Content here is licensed CC BY 4.0 — share and adapt freely, including commercially, with credit to "FAPD — Free Agentic Publication Digester". Quoted official government text within it is public domain (17 U.S.C. § 105) and needs no permission at all. The attribution rule mirrors our citation ethic: for factual claims, cite the underlying official source each item links to; cite FAPD for the aggregation and summaries. The pipeline's code is Apache-2.0.