A record you can inspect.

Read the limits alongside the results.

A recent refresh is not complete coverage. The latest date varies by chamber. A gap in our records does not mean a representative was inactive, absent, or silent. Attendance figures describe only the roll calls we hold.

Topics help you find evidence. Most bills are tagged with up to three topics by two independent AI taggers reading the bill title and summary, with a third deciding where they disagree; the rest are matched on words in the title and summary. Tags can miss relevant votes or include imperfect matches. They do not label a vote as supporting or opposing your beliefs. Search alternative terms and open the official record.

Vote summaries. Congress, New York and some states publish their own summaries, which we show as “official.” Where a state publishes none, the summary is AI-written from the official bill page (digest, fiscal note or bill text) and checked against that page by a second AI pass; it is labeled as AI. One summary covers a bill, so on a bill that changed during the session it can describe a different version than the one a vote was on. The official record link shows exactly what was voted. Procedural and amendment votes may carry an AI-written line explaining what a yes vote did, checked the same way.

A vote is on a specific question. Procedural votes, amendments, confirmations, and final passage have different meanings. A bill passing one chamber does not by itself make it law. Large bills can contain several unrelated policies.

Candidate research is a separate collection. Profile records and selected key votes may cover other dates and are not counted in this snapshot inventory. Research depth varies. Check each race’s research date, notes, and cited sources.

State legislative candidate lists. Candidate lists for 5,382 state legislative contests in 45 states come from Ballotpedia's statewide general-election tables, generated Oct 19, 2026. They list names and parties only. Ballotpedia notes its lists may be incomplete, write-in candidates are not marked, and seats decided in the primary (Hawaii, Oklahoma) are omitted. New York is researched separately, and a state or chamber with no seat up this year, or not in the file, shows no claim.

We welcome corrections. Report a problem with the page URL and a primary source. Our research methodology explains source selection and AI-assisted summaries.