API Reference

REST API endpoints for the keyes.ai Memory-as-a-Service.

Quick Start

All requests require an API key in the Authorization header:

curl -X POST https://keyesmemory.com/v1/memory/search \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer mol_xxxx" \
  -d '{"text": "hello world", "top_k": 5}'

Endpoints

POST/v1/memory/searchL1

Semantic vector search. Send text (auto-embedded) or a pre-computed vector.

{
  "text": "meeting about Q3 budget",
  "collection": "work_memories",
  "top_k": 10
}

POST/v1/memory/filterL2

Filtered semantic search. Combine vector search with metadata predicates.

{
  "text": "budget projections",
  "filter": { "field": "category", "op": "eq", "value": "finance" },
  "top_k": 5
}

POST/v1/memory/storeWrite

Store a memory. Text is auto-embedded. Metadata is stored alongside the vector.

{
  "text": "Q3 budget approved at $2.4M",
  "metadata": { "category": "finance", "date": "2026-03-04" },
  "collection": "work_memories"
}

DELETE/v1/memory/:idWrite

Delete a memory point by its ID.

POST/v1/memory/joinL3

SQL→Vector join. Run a SQL query to generate a filter, then search with it.

{
  "sql_query": "SELECT id FROM users WHERE tier = 'premium'",
  "text": "product feedback",
  "top_k": 10
}

GET/v1/memory/statusFree

Health check. Returns service status.

POST/v1/license/checkFree

Validate API key and return tier information.