Federal Register Scraper

The Federal Register scraper for compliance, policy and lobbying teams: search the US Federal Register and get rules, proposed rules, notices and presidential documents as JSON — with agencies, effective dates, public comment deadlines with days remaining, docket ids and the sections of federal regulation affected.

Reads the Register's own public API. No key, no login.

The comment deadline is the field with consequences

A proposed rule is open for public comment until a date, and after that date the opportunity to influence it is gone. That is the one piece of this data people actually act on, and it is one of the fields the API leaves out unless you ask for it by name.

Every row carries comments_close_on, plus two derived fields:

• comment_days_remaining — days until the deadline, negative once it has passed. A closed deadline being visible matters as much as an open one.
• comments_open — whether the window is still open today.

Set comments_open_only and you get exactly the documents you can still respond to. Measured on 6 October 2026, that filter returned 179 proposed rules still open for comment. The run summary adds openForComment for whatever you actually received, so a daily run of this filter is a comment calendar rather than a reading list.

Ask for your fields or you will not get them

The API returns a small default set and silently omits comments_close_on, effective_on, docket_ids, regulation_id_numbers, significant, cfr_references and more. Nothing in the response indicates anything was left out, so a scraper that does not name its fields concludes those values do not exist for these documents.

This Actor always requests the full set.

Output: one row per document, with a real example

One row per document, all keys present in every row:

• Identity: document_number, title, type, citation
• Content: abstract, action, topics, dates_text
• Who: agencies, president
• When: publication_date, effective_on, comments_close_on, comment_days_remaining, comments_open
• Regulatory: significant, docket_ids, regulation_id_numbers, cfr_references
• Where to read it: html_url, pdf_url, raw_text_url
• Extent: page_length, start_page, end_page
• Provenance: document_id, fetched_at, source, country, position

cfr_references is returned readable — 40 CFR 60 rather than a nested object — because that is the form a compliance team recognises.

raw_text_url is the plain-text version, which is the one to feed to anything doing full-text analysis rather than the PDF.

page_length is a rough indication of how substantial a document is: a three-page notice and a three-hundred-page rule are different animals.

Beside the rows, every run writes a summary (documentsReturned, reportedCount, countNote, pagesFetched, typeBreakdown, openForComment, withEffectiveDate, fetchedInSeconds, errorCount) and, when something went wrong, an errors list with a code and a redacted message. Both are readable from the key-value store of the run; the dataset holds the rows.

A real run — search: artificial intelligence, limit: 5, 6 October 2026 — reported 1,587 matches, delivered 5 rows (2 proposed rules, 1 notice, 1 presidential document, 1 rule), 3 of them still open for comment. One of those rows, exactly as delivered:

{
  "document_number": "2026-20474",
  "title": "Investment Adviser Performance-Based Compensation Modernization",
  "type": "Proposed Rule",
  "abstract": "The Securities and Exchange Commission (the \"Commission\") is proposing to amend the rule under the Investment Advisers Act of 1940 that provides an exemption from the statutory prohibition on registered investment advisers receiving compensation on the basis of a share of capital gains in or capital appreciation of an advisory client's account. Specifically, the proposed amendments would expand the ability of investment advisers to receive this compensation from clients that are registered management investment companies and business development companies (collectively, \"regulated funds\"), subject to certain conditions. The proposal would relatedly amend certain regulated fund registration and reporting forms to require separate disclosure of all performance-based compensation paid by regulated funds to their investment adviser. The proposed rule amendments would also allow investment advisers to receive this compensation from additional clients by revising the rule's \"qualified client\" definition to include investors that meet the \"accredited investor\" definition in Regulation D under the Securities Act of 1933. The proposal would relatedly make conforming amendments to certain other rules under the Investment Advisers Act of 1940 whose provisions reference the \"qualified client\" definition.",
  "action": "Proposed rule.",
  "agencies": [
    "Securities and Exchange Commission"
  ],
  "publication_date": "2026-10-06",
  "effective_on": "2026-10-06",
  "comments_close_on": "2026-12-07",
  "comment_days_remaining": 61,
  "comments_open": true,
  "significant": null,
  "docket_ids": [
    "Release Nos. 33-11443",
    "34-106533",
    "IA-7022",
    "IC-36350",
    "File No. S7-2026-28"
  ],
  "regulation_id_numbers": [
    "3235-AN59"
  ],
  "topics": [
    "Brokers",
    "Fraud",
    "Investment companies",
    "Reporting and recordkeeping requirements",
    "Securities"
  ],
  "cfr_references": [
    "17 CFR 239",
    "17 CFR 249",
    "17 CFR 274",
    "17 CFR 275"
  ],
  "citation": "91 FR 63676",
  "page_length": 70,
  "start_page": 63676,
  "end_page": 63745,
  "president": "Donald Trump",
  "dates_text": "This release was published in the Federal Register on October 6, 2026. Comments should be received on or before December 7, 2026.",
  "html_url": "https://www.federalregister.gov/documents/2026/10/06/2026-20474/investment-adviser-performance-based-compensation-modernization",
  "pdf_url": "https://www.govinfo.gov/content/pkg/FR-2026-10-06/pdf/2026-20474.pdf",
  "raw_text_url": "https://www.federalregister.gov/documents/full_text/text/2026/10/06/2026-20474.txt",
  "document_id": "2026-20474",
  "fetched_at": "2026-10-06T21:02:02.930Z",
  "query": "artificial intelligence",
  "source": "federalregister-scraper",
  "country": "US",
  "position": 1
}

Note the five fields the Register's default response never gave you: the deadline with its days left, the docket ids, the regulation id numbers, the CFR references as plain text, and significant — null here, because absent is not the same as "not significant".

A note on the reported count

The Register caps its reported count at 10,000. Measured on 6 October 2026, an unfiltered list reported exactly that number, which is a display ceiling rather than a measurement.

Narrower queries return real numbers: the same day, artificial intelligence reported 1,587 matches and open proposed rules 179. So the count is trustworthy below the ceiling and a floor at it. The run summary sets countNote only when you are sitting on the ceiling, rather than letting you treat 10,000 as a total.

Federal Register scraper input: filters, limits and defaults

Field	Type	Default
search	free-text term	—
doc_type	rule, proposed_rule, notice, presidential_document	any
agency	agency slug, e.g. environmental-protection-agency	—
comments_open_only	only documents still accepting comment	false
date_from, date_to	YYYY-MM-DD	—
order	newest, oldest, relevance	newest
limit	1-5000	100
retries	1-6	3

Every field is optional: an empty input returns the newest documents of the whole Register. Agency names are slugified for you, so Environmental Protection Agency and environmental-protection-agency both work.

Use cases for a Federal Register scraper

• Regulatory monitoring. Filter by agency and comments_open_only, run daily, and you have every rule your industry can still respond to, with the deadline attached.
• Compliance calendars. effective_on is when a rule bites. cfr_references says which parts of the code change.
• Policy research. Term search across decades, with raw_text_url for the documents you want to read in full.
• Competitive and lobbying intelligence. docket_ids link Register documents to the dockets where comments are filed.
• Executive action tracking. doc_type: presidential_document returns executive orders and proclamations; president is filled on every row, so you can also group any result set by the administration in office.
• Datasets for analysis. abstract, topics, citation and page_length are flat columns, ready for a spreadsheet, a database or a text-mining job.

Federal Register scraper vs the Register's own search and API

The Register's website is built for reading one query at a time. It is good at that, and it is free, but a recurring watch — a list of agencies, a deadline calendar, a dataset of thousands of documents — means clicking, exporting and reconciling whatever the page hands you.

The Register's own API (what this Actor calls) is also free and needs no key, but it ships a small default field set, pages 100 documents at a time, retries only what you build yourself, and knows nothing about days remaining: comment_days_remaining and comments_open are computed here, not requested there. Its reported count also stops at 10,000, as described above.

This Actor does the field selection, the paging, the retries and the deadline arithmetic for you, and returns one flat, consistent row per document with the same field names on every run. You can then export the dataset as JSON, CSV, Excel or HTML from the run page, or read it through the Apify API in Python, JavaScript or any other language. Scheduling, storage and integrations come from the Apify platform, so a daily regulatory watch needs no server of your own.

There are other Federal Register actors in the Store — a search for "Federal Register Scraper" returns hundreds of them — so compare on what the row actually contains: a dated comment deadline, docket ids, CFR references as readable text, and a price you can predict.

Notes

significant is the Register's own flag and is frequently null rather than false; absent is not the same as "not significant" and is returned as-is.

An empty search is a successful empty run: the Register answers a query with no matches with HTTP 200 and a count of zero, the run records a no_results error, and no document is charged. A rejected search condition is reported the same way rather than retried, since retrying a bad condition cannot help.

Coverage is the US Federal Register only. State registers are published separately and are not in this data.

Finding an agency slug

Agency slugs are the agency name lower-cased with hyphens: environmental-protection-agency, food-and-drug-administration, securities-and-exchange-commission, federal-communications-commission. The Actor slugifies whatever you type, so the plain name works too.

A document is frequently issued by several agencies at once, and agencies returns all of them. Filtering by one agency returns documents where it is any of the issuers, not only the lead, which is usually what you want and occasionally a surprise.

What this Actor does not do

No document full text. raw_text_url gives you the address of the plain text, and fetching it is a separate job with a different cost profile: some rules run to hundreds of pages.

No public comments. The Register publishes the documents and the deadlines; the comments themselves live on Regulations.gov, which is a different API.

No historical amendments. A rule's row describes the rule as published. How it was later amended lives in the eCFR, not here.

Pricing and free-plan limits

Pricing is unchanged: $0.001 per run start, plus $0.001 per document delivered. A run of 100 documents costs $0.101 in Actor event charges; 1,000 documents cost $1.001. The start fee applies even when the search returns nothing or the input is rejected, so an empty result produces no document charge.

Non-paying Apify users receive at most 10 documents per run: limit is capped at 10 before the fetch, so you get the first ten rows of exactly the same query. Paying users can request up to 5000 documents per run.

For comparison: a Google result for this query advertised pricing from $19.00 per 1,000 when we measured on 6 October 2026. The Register's own API is free, and this Actor's document price is what you pay for the fields, the paging and the deadlines.

FAQ

Do I need an API key?

No. The Actor reads the Register's publicly published API. There is no key, no login and no account to create at the source.

Does an empty search fail the run?

No. No matches produce a successful empty run with an explanatory no_results record. The run-start fee still applies, and no document is charged.

How many documents can one run return?

limit accepts 1 to 5000. Users on a free Apify plan receive at most 10 documents per run; upgrading removes that cap. Documents are deduplicated on document_number, and position runs 1..n across the delivered rows.

Can I track comment deadlines across agencies at once?

Yes. Leave agency empty, set comments_open_only: true, and every document in the result is one you can still respond to, with comments_close_on and comment_days_remaining on the row. Schedule that run once a day and the deadline never leaves your calendar.

Do I get the full text of a document?

No. raw_text_url points at the plain text of each document; fetching hundreds of pages is a job with a different cost profile. This Actor returns the metadata, the abstract (up to 3000 characters as the source provides it) and the links.

Can I call this Federal Register API from Python?

Yes. Start the Actor with the Apify API client for Python or JavaScript, or with a plain HTTP request, pass the same input fields as in the form, and read the dataset items when the run finishes. The document_number is the stable key for joining runs.

Which parts of federal regulation are covered?

Only the Federal Register: rules, proposed rules, notices and presidential documents from the US federal government. The Code of Federal Regulations, state registers and the comment transcripts on Regulations.gov are separate publications with separate sources.