Full-text search

GET /api/v1/search meetings:read

Searches minutes, meeting titles and transcripts. A hit carries a highlighted excerpt (<mark>) and where it was found (protocol | title | transcript). At most one hit per meeting.

Query parameters

Name	Type	Required	Description
q	string	yes	At least 2 characters.
limit	integer	—	Between 1 and 100. Default 50.

Response 200

The hits, best first. No pagination — limit is the upper bound.

Example response

{
  "query": "quote",
  "data": [
    {
      "meeting_id": "8f14e45f-ceea-467a-9e2b-1d3f5c0a7b21",
      "title": "Quarterly review",
      "platform": "string",
      "starts_at": "2026-08-04T08:00:00Z",
      "created_at": "2026-08-01T09:12:44Z",
      "rank": 0.42,
      "snippet": "we send the <mark>quote</mark> by Friday",
      "matched_in": "protocol"
    }
  ]
}

Field	Type	Required	Description
query	string	yes	The search term, trimmed.
data	object[]	yes	The hits, best first.
data[].meeting_id	uuid	yes	The meeting the hit was found in.
data[].title	string	yes	Subject, as the calendar or the upload spelled it.
data[].platform	string	—	Where that meeting took place.
data[].starts_at	date-time | null	—	Start of the appointment. null for an upload, which has none.
data[].created_at	date-time	—	When the meeting was first recorded in Protoki.
data[].rank	number	yes	Hit score. A title hit always carries 0.35.
data[].snippet	string	yes	Excerpt with <mark>…</mark> around the place found. This contains HTML — handle it before you display it.
data[].matched_in	string	yes	Which part of the meeting the hit came from. Values: protocol, title, transcript.

Other answers

Code	Meaning
400	Invalid parameter — the text names the field and the shape expected.
401	No key, or a key that is not valid.
403	The key is missing the scope this endpoint needs.
503	The endpoint does not exist on this deployment yet, because a database change has not been applied. A temporary state, not a mistake in your request.