Semantic search

Search memories by meaning, not exact words. Use OpenAI embeddings to find relevant knowledge even when you use different terms.

Semantic search

POST/api/v1/search/semantic

bash

curl -s -X POST "$SOLUCORTEX_URL/api/v1/search/semantic" \
  -H "Authorization: Bearer $SOLUCORTEX_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Accept: application/json" \
  -d '{
    "project_id": "'"$SOLUCORTEX_PROJECT_ID"'",
    "query": "how is authentication implemented?",
    "limit": 10
  }'

Parameters

Field	Type	Required	Description
`project_id`	UUID	✅	Project UUID
`query`	string	✅	Question or description in natural language
`limit`	int	—	Number of results. Default: 10, max: 50
`type`	string	—	Filter by memory type
`threshold`	float	—	Minimum similarity 0–1. Default: 0.5

Response

json

{
  "query": "how is authentication implemented?",
  "results": [
    {
      "id": "019e...",
      "type": "decision",
      "title": "API keys with SHA-256 hash",
      "content": "Keys are stored as hashes...",
      "importance": 9,
      "score": 0.94,
      "created_at": "2026-05-13T10:00:00Z"
    },
    {
      "id": "019f...",
      "type": "architecture",
      "title": "Middleware api.key validates Bearer token",
      "score": 0.87
    }
  ]
}

score is cosine similarity (0–1). Values above 0.8 are highly relevant.