embedding.pending · Umwelt docs

Reference/Queries/embedding.pending

embedding.pendingquerysince 0.1

The searchable documents whose vector in a space is missing or out of date, with the exact text to embed.

Command lineTypeScript.NET

Request

umwelt call story.umw embedding.pending '{
  "projectId": "ball",
  "spaceId": "demo",
  "docTypes": ["record"],
  "page": { "limit": 2 }
}'

Request

const result = await client.embeddingPending({
  projectId: 'ball',
  spaceId: 'demo',
  docTypes: ['record'],
  page: { limit: 2 },
});

Request

var result = engine.EmbeddingPending(new PendingEmbeddingsInput
{
    ProjectId = "ball",
    SpaceId = "demo",
    DocTypes = new() { "record" },
    Page = new Page { Limit = 2 },
});

Response

{
  "items": [
    {
      "stale": false,
      "docId": "maid-staff",
      "docType": "record",
      "text": "maid umwelt member of household",
      "textHash": "04d718cba11bc4995d03fbe82412f2bd2d6156acb6dc27042c297ed94a4d01f2"
    },
    {
      "stale": false,
      "docId": "butler-staff",
      "docType": "record",
      "text": "butler umwelt member of household",
      "textHash": "5502f404e90632e72e7395a6f6d15dc5837dee8e9dbbfea82020527e4fffea1c"
    }
  ],
  "nextCursor": "3132",
  "resumeCursor": "3132"
}

Payload

Field	Type	Description
`projectId`required	string	The project to read.
`spaceId`required	string	The embedding space.
`docTypes`	string[]	`record` and/or `event`. Empty: both.
`page`	Pagedefault {"limit":100}	How many documents to return (`limit`) and where to continue (`cursor`: `nextCursor` to page, `resumeCursor` to see only newer documents).
`timelineId`	stringdefault "main"	The timeline to read.

Result

Field	Type	Description
`items`always	PendingEmbedding[]	Documents in the order they were indexed, each with its `text`, `textHash` (SHA-256) and whether a stored vector is `stale`.
`nextCursor`	string	More pending documents follow; pass it as `page.cursor`.
`resumeCursor`always	string	Where a later call can start to see only documents written after this one. Starting without a cursor checks every document again.

Rules

Embed exactly this text. Store the vector with the textHash it came from. A vector whose hash no longer matches stays pending.

The engine never runs a model. Your model makes the vectors. The TypeScript and .NET clients can call it for you.

Errors

VALIDATION_FAILEDThe payload is malformed, or a doc type is unknown.

NOT_FOUNDThe project, space or timeline does not exist.

INVALID_STATEThe timeline read is a draft that was promoted. details.promotedInto names the timeline to read instead.

HANDSHAKE_REQUIREDOver the protocol, hello was not sent first.

Every code is listed in Errors.

Payload

Result

Rules

Errors

See also