GraphQL

POST https://www.chicagopolitics.org/api/graphql with {"query": "...", "variables": {...}}. One typed graph of everything: ask for exactly the fields you need, across related records, in one request. This is usually the best fit for AI agents. The schema needs no key, and introspection works.

Your first query

curl https://www.chicagopolitics.org/api/graphql \
  -H "Authorization: Bearer $CHICAGO_POLITICS_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "{ ward(number: 4) { member { name } } }"}'

{"data":{"ward":{"member":{"name":"Lamont J. Robinson"}}}}

That is 58 bytes, against about 4.8 KB for the full member record.

A ward's member, their latest no votes, and the legislation and meeting behind each:

{
  ward(number: 15) {
    member {
      name
      votes(vote: NAY, first: 5) {
        totalCount
        nodes {
          rollCall {
            title date tally { yea nay }
            legislation { record aiSummary { summary } }
            meeting { body start }
          }
        }
      }
    }
  }
}

Reference

Root fields: member, members, ward, wardAt(latitude, longitude), committee, committees, legislation(id or record), searchLegislation, rollCall, rollCalls, meeting, meetings, insights, money, donations, lobbyistContributions, lobbyistGifts, searchDocuments (semantic), answer (AI, cited), me; mutation contributeData. Members also have money, donations, and spending.
Lists page with first (max 200) and after (from pageInfo.endCursor). Every object has a url to cite.
One query counts as one request; answer also uses an AI call. Limits per query: depth 10, 250 fields, one answer, three searchDocuments.
Fields are added without breaking existing queries; anything retired is marked @deprecated first.

GraphQL

Your first query

Related records in one request

Reference