Skip to main content
Two fields answer “can I rely on this section today”, and they answer different questions. actStatus is the publisher’s own word for the section, stored as we read it. goodLawStatus is our verdict, derived from that word plus whether the jurisdiction publishes a repeal signal we trust. Both appear on POST /us/statutes/search, GET /us/statutes/section/{actId}, and POST /us/statutes/sections. actStatus is also a search filter, which is covered below.

​
actStatus

The raw status of the section, straight from the source structure. It is the only one of the two that is indexable, so it is what you filter on. Operative Dormant: it exists, but has no operative force right now Dead: it is not law, or is no longer at this citation A dead or dormant status is a real answer about the law, not a data quality flag. Several of these exist because one jurisdiction uses its own verb: recompiled is New Mexico’s, not_funded is the District of Columbia’s, relocated is Alaska’s, terminated and vetoed are Wyoming’s. We store the publisher’s word rather than flattening it, so a section whose entire printed body is Recompiled. is not served as live law that happens to say almost nothing.

​
goodLawStatus

A derived, conservative verdict computed on top of actStatus and jurisdiction-specific repeal signals.
goodLawStatus is deliberately conservative. Some jurisdictions and some publishers do not publish a reliable repeal signal, so their sections resolve to unknown rather than over-claiming good_law.Treat unknown as “verify against the official source,” not as “current.” The field is null when currency checking is disabled.

​
Filtering on status

There are two filters on POST /us/statutes/search, and they are not inverses of each other in the way the names suggest. excludeRepealed: true drops everything whose own status says it is no longer operative: the dead and dormant vocabularies above, in full.
excludeRepealed removes what we KNOW is dead. It does not promise the remainder is good law.A section survives that filter when its status is in_force OR when we hold no trustworthy repeal signal for its jurisdiction, and those two are not the same thing. Read goodLawStatus on each surviving result to tell them apart: good_law is checked, unknown is unchecked.
actStatus on the request scopes positively to one or more raw statuses, as a string or a list. It is what you want for a compliance diff that asks what was LOST rather than what remains.
The values you may pass are the values a result carries, listed above. An unrecognized one is rejected with a 422 rather than silently matching nothing, so read the enum on the endpoint’s OpenAPI schema if you want the machine-readable set. Combining a dead status here with excludeRepealed: true contradicts itself and is rejected with a 422 as well.

​
Currency and move pointers

  • currencyNote: the source’s own “current through …” line, when the publisher provides one. Use it to show how fresh the text is, rather than a bare fetch date.
  • renumberedTo: when actStatus is renumbered, a pointer to the new location.
  • transferredTo: when actStatus is transferred, a pointer to the new code or title.
Follow renumberedTo / transferredTo to fetch the section’s current home.

​
Using it

Gate your product on goodLawStatus, surface currencyNote, and follow a move pointer when a section has been relocated.

​
Cross-references between statutes and regulations

USC and CFR sections carry each other’s citations where the source material draws the link, so you can walk from a statute to the regulations that implement it, or from a regulation back to the law that authorizes it, without a separate search.
  • statutoryAuthority (CFR results): the USC citation(s) the regulation cites as its authorizing statute.
  • implementingRegulations (USC results): the reverse index, which CFR sections cite this statute as their authority.
  • crossReferencesCfr / crossReferencesUsc: parsed cross-references to the other corpus found in the section’s own text, as a list of section strings (e.g. ['1.6011-4']). A section referencing itself is represented as 'self:<section>' rather than omitted.
All four are nullable and populated only where the underlying source material actually draws the connection. Absence does not mean no relationship exists, only that it was not extractable from the text.

​
Federal Register tracking fields

FEDERAL_REGISTER results (final and proposed rules) carry rulemaking-lifecycle fields not present on statute or regulation results, useful for tracking a rule from proposal through effect: These are populated on GET /us/statutes/section/{actId} (not on search results) for sections under corpusType=FEDERAL_REGISTER.

​
Why the verdict differs by corpus

goodLawStatus is not uniform inside a corpusType. It is decided by the PUBLISHER, because only a publisher that records a retirement lets us detect one. The rule is an allow-list: a source earns a good-law verdict only after its own rescission signal has been measured, and a new source is unknown until then rather than certified by default. Read these if you depend on one of the corpora named.
goodLawStatus under corpusType=STATE_AGENCY_GUIDANCE is decided by the publisher that issued the document, and a state can hold two publishers with different answers.Arizona is the worked example. Its insurance regulator marks rescissions diligently, so source=state_insurance_bulletin earns good_law: 15 of its 199 bulletins are served with a dead actStatus on the strength of the Department’s own withdrawal statements. Its financial-institutions regulator, which shares the same index, has rescinded nothing we hold: 0 of 35. So source=state_financial_bulletin is served unknown in both Arizona and Oregon, and will be until a regulator actually withdraws something and we store it.Same state, same corpusType, same actStatus, two different verdicts, because they are two different publishers. Read unknown here as “we make no claim”, not as “probably still current”.
Every section under corpusType=STATE_AG_OPINION is served with goodLawStatus: "unknown", in every jurisdiction, and it is a construction rather than a gap.An AG opinion is not amended. It is withdrawn, overruled, modified, or left standing — and almost no publisher records which. Surveying all 52 jurisdictions, exactly three print a supersession signal of any kind: Georgia publishes an “Obsolete and Modified Opinions” ledger, Ohio publishes an Overruled Opinions list, and Wyoming annotates a withdrawn opinion in its own index with the case that overtook it. The other 49 publish nothing, so for them an opinion issued in 1985 and an opinion issued last month are indistinguishable on currency.None of the three has yet been measured against the criteria we require before certifying a publisher, so the allow-list is empty and nothing in this corpus is promoted.Ohio’s list has now been measured, and it changes less than its title suggests. Parsed in full on 2026-09-12: 584 rows, of which only 47 name an opinion we serve and only 20 of those 47 say the opinion is dead — 1.42% of Ohio’s served archive. The other 27 say followed by, approved, clarified or modified in part, which mean the opinion still stands, so a list headed “Overruled Opinions” is 43% not overrulings. Most of the ledger, 536 of 584 rows, is about Ohio’s 1920-1992 era, which we do not serve.So even the best supersession signal in this corpus reaches a small, mostly-historical fraction of it, and reading a row requires reading its verb. That is why the measurement does not promote Ohio out of unknown.Read unknown here as “we make no claim about whether the Attorney General still stands behind this”, not as “probably still current”. And note the separate point that an AG opinion is persuasive authority, not binding law even when it is current: see Corpus types.
Every corpus under corpusType=AGENCY_ADJUDICATION except one is served with goodLawStatus: "unknown", and that is a construction rather than a gap.We serve the decision exactly as the body issued it, and we have NOT verified that it is still good law. Read unknown as “we make no claim”, not as “still good”.The reason is the publisher, not the pipeline. An adjudication is superseded by ANOTHER DECISION, not by an amendment to itself, so the document never changes when it stops being law and none of the ordinary currency instruments fire. No publisher in this family ships a structured overruling or supersession field.nlrb_board_decision is the clearest case, and it is deliberately not on the allow-list. Board doctrine swings hard with Board composition, so an overruling is the normal course rather than an edge case, and the NLRB publishes nothing machine-readable that records one. Its CITENET Classified Index is a subject digest, not a citator: it carries an overruling only as editorial prose on the decision doing the overruling, with no reverse edge on the decision that was overruled. Two independent measurements, the citenet probe and a supersession sweep over the landed corpus, both returned no signal. So we serve the Board’s text faithfully and refuse to assert what a later Board has done to it.mspb_precedential, mspb_nonprecedential, olc_opinion, sec_commission_opinion, bia_precedent, ag_precedent and immigration_admin_precedent are unknown for the same reason.
Do not treat goodLawStatus: "unknown" under this corpusType as a soft yes. On a body whose doctrine turns over with its membership, a decision served as issued may have been overruled the week after it was decided, and nothing in this API will tell you. If your use of an adjudication depends on it still being good law, check the body’s own subsequent output.
ftc_administrative_decision is unknown for a related but distinct reason, because it is the one place where the document tells you something the API cannot.Since 1995 the Commission’s standard consent order terminates 20 years after it issues, and many orders carry an explicit sunset paragraph in their own text. So a 2001 consent order is very likely spent today.Nothing in the FTC’s published metadata says so. The sunset is prose inside the order, with its own conditions, exceptions and tolling, and we deliberately do not compute an expiry from issued_date + 20 years. Inferring a legal conclusion from arithmetic on a date is exactly the kind of invented fact this API refuses to serve. Read the order’s own terms.The Commission’s Case Status facet does not close that gap either, and should not be read as if it did: it says a MATTER is closed, not that the order which closed it stopped binding. It is held with the document and is never mapped to a dead actStatus, and it is not exposed as a response field.
ftc_hsr_interpretation is unknown, and it is worth saying what that does and does not mean here, because this source carries real supersession data.The Commission published Formal Interpretation No. 13 to supersede No. 5, and says so in the document’s own identifier. It publishes the explanatory material for No. 15 in three dated versions, 1998, July 1999 and March 2001. Both relations are served, as supersedes and supersededBy carrying act ids, so you can walk from any version of an interpretation to the one that replaced it.What we do not do is convert that into a currency verdict. Every member of those chains is served with actStatus: "in_force", including the interpretations that were superseded.The reason is the size of the series. There are 20 formal interpretations, so one wrong link is 5% of the corpus, and a dead actStatus removes a document from retrieval altogether. The earlier interpretation is also usually exactly what a reader of the later one needs to see. So we serve the sequence and let you judge it, rather than asserting a conclusion the Commission never printed.The informal half carries no supersession data at all, and is in any case staff views rather than Commission action: the Premerger Notification Office answering a practitioner’s question, not binding on the Commission.⚠️ effectiveDate is absent on 13 of the 20 formal interpretations. The FTC publishes no date field for that series, and those 13 print dates only inside their own argument, where the document’s own date cannot be distinguished from the dates of the rules and correspondence it discusses. We leave the field empty rather than guess it.
ftc_advocacy_filing is unknown for a reason that is different in kind from every other source on this page: currency is close to the wrong question.An advocacy filing is the Commission or its staff writing to a state legislature, a state supreme court, a governor or another federal agency about what a proposed measure would do to competition. It binds nobody. It decides no case, it construes no FTC rule, and the recipient is free to ignore it. There is no status it could lose, because it never had the force these other sources have.It is held as its own source, with its own forum value (ftc_advocacy), so you can scope it in or out of an AGENCY_GUIDANCE query in one parameter, and every document carries a coverageNote saying this in its own payload.
Most of the corpus derives goodLawStatus from a repeal signal in the text. One source instead carries the issuing agency’s own published lifecycle status, and it behaves differently from everything around it.The Consumer Financial Protection Bureau assigns every enforcement action one of three statuses and publishes the definitions itself:Two consequences you should plan for.185 of the 386 actions carry the dead status, so they are excluded from ordinary search results. That is close to half the source, and it is correct: an order that no longer binds anyone should not be returned as live law. Query them explicitly with actStatus: "terminated" if you need the historical record.That status bundles three different endings and the Bureau does not separate them. A matter it lost or withdrew is not the same event as an order that ran its term, and no field distinguishes them. Each record also carries the Bureau’s raw status string verbatim in history, and you should read that rather than inferring an outcome from actStatus alone.This is why CFPB enforcement actions carry a real verdict while every other source under AGENCY_ADJUDICATION carries unknown. Filter on source if you need one behavior or the other.
supersessionActions records what a document does to EARLIER documents, keyed by the issuer’s own verb: modifies, obsoletes, amplifies, clarifies, rescinds, each mapping to the documents affected.supersedes and supersededBy cover only the strongest of those verbs. Without supersessionActions a caller cannot tell a modified revenue procedure from an untouched one, nor see which memorandum rescinded which.rescindedOn carries the date, where the issuer states one.

Section Identifiers

The actId handle and the search then section then body workflow.

Response Fields

Which fields each corpus populates, and point-in-time text with asOf.

Grounding LLMs

Cite only good-law sections and link back to the official source.

Coverage

Which corpora are ingested and how often each refreshes, per jurisdiction.
Last modified on September 12, 2026