POST /v1/recall

Retrieve relevant memories using a natural language query. CortexDB finds the most relevant context and returns it in a format ready for use with LLMs.

Request

POST /v1/recall
Content-Type: application/json
Authorization: Bearer <api-key>

Body

{
  "query": "What database are we using for the payments service?",
  "tenant_id": "my-app"
}

Parameters

| Field | Type | Required | Description | |---|---|---|---| | query | string | Yes | Natural language query | | tenant_id | string | No | Tenant identifier (default: "default") | | max_tokens | integer | No | Maximum token budget for response (default: 4096) | | min_confidence | float | No | Minimum confidence threshold (default: 0.0) |

Response

Status: 200 OK

Python SDK Response

{
  "context": "Chose CockroachDB for the payments service. Rationale: multi-region support, PostgreSQL compatibility. Migration completed March 2026.",
  "confidence": 0.94,
  "latency_ms": 87
}

| Field | Type | Description | |---|---|---| | context | string | Synthesized context from matching memories | | confidence | float | Confidence score (0-1) | | latency_ms | integer | Server-side processing time |

TypeScript SDK Response

{
  "matches": [
    {
      "content": "Chose CockroachDB for the payments service. Rationale: multi-region support, PostgreSQL compatibility.",
      "confidence": 0.94,
      "episodeId": "ep_a1b2c3d4",
      "metadata": {}
    }
  ],
  "total": 1
}

| Field | Type | Description | |---|---|---| | matches | array | Ranked list of matching memories | | matches[].content | string | Memory content | | matches[].confidence | float | Relevance score (0-1) | | matches[].episodeId | string | Episode ID | | matches[].metadata | object | Episode metadata | | total | integer | Total matches |

Example

curl -X POST https://api.cortexdb.io/v1/recall \
  -H "Authorization: Bearer your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "What database are we using for payments?",
    "tenant_id": "my-app"
  }'