recall

Recall the k most-similar memory cells to a query at layer L0, scoped to the caller. The server embeds the query text and returns distance-ranked node ids, nearest first. Dispatches to the recall internal service method. For an explicit semantic layer, use query instead.

Parameters

Name	Type	Required	Default	Description
`text`	string	Yes	—	The query text, embedded server-side.
`k`	integer	No	`10`	Number of neighbours to return (minimum `1`).
`filter`	object	No	—	A convenience `NodeFilter`, AND-ed after the caller’s access scope.

Returns

{
  "nodes": [
    { "node_id": "<uuid>", "distance": <number> }
  ]
}

Returned as the structuredContent payload. Results are ordered nearest first (smallest distance).

Over MCP, recall returns only {node_id, distance} pairs — not hydrated cell content. To read a cell’s content, use a hydrating surface such as the REST API, which returns full nodes.

Example

Request:

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "recall",
    "arguments": { "text": "which vector store did we pick?", "k": 3 }
  }
}

Response:

{
  "jsonrpc": "2.0",
  "id": 1,
  "result": {
    "content": [
      { "type": "text", "text": "{\"nodes\":[{\"node_id\":\"6f9619ff-8b86-d011-b42d-00cf4fc964ff\",\"distance\":0.12}]}" }
    ],
    "structuredContent": {
      "nodes": [
        { "node_id": "6f9619ff-8b86-d011-b42d-00cf4fc964ff", "distance": 0.12 }
      ]
    },
    "isError": false
  }
}

Annotations

The tool’s MCP behaviour hints, as emitted in tools/list (advisory display metadata — see the annotations contract):

`readOnlyHint`	`destructiveHint`	`idempotentHint`	`openWorldHint`
`true`	`false`	`true`	`false`

A pure vector read — writes nothing.

query for layered search across L0/L1/L2.
The HNSW mesh for how nearest-neighbour search works.

Overview

Tools

Parameters

Returns

Example

Annotations

​Parameters

​Returns

​Example

​Annotations

​Related

Parameters

Returns

Example

Annotations

Related