Prepare Context

Endpoint

POST /v1/memory/prepare

Permission: Read

Request Body

{
  "namespace": "my-app",
  "entity": "user-123",
  "messages": [
    { "role": "user", "content": "Help me set up my workspace" }
  ],
  "maxCandidates": 20,
  "maxTokens": 500,
  "category": "preference"
}

Field	Type	Required	Default	Description
`namespace`	`string`	✅	—	Memory namespace
`entity`	`string`	✅	—	Entity identifier
`messages`	`Message[]`	✅	—	Conversation messages for context analysis
`maxCandidates`	`number`	—	`20`	Max memories to consider
`maxTokens`	`number`	—	—	Guide the output token budget
`category`	`string`	—	—	Filter candidates by category

Message Format

{
  "role": "user" | "assistant" | "system",
  "content": "string"
}

Response

{
  "context": "The user prefers dark mode and uses TypeScript for all projects. They work as a software engineer and their team uses React with Next.js.",
  "memoriesUsed": [
    "550e8400-e29b-41d4-a716-446655440000",
    "550e8400-e29b-41d4-a716-446655440001"
  ],
  "candidatesConsidered": 15,
  "tokenCount": 127
}

Field	Type	Description
`context`	`string`	Synthesized briefing paragraph — ready to inject into a system prompt
`memoriesUsed`	`string[]`	IDs of memories used in the synthesis
`candidatesConsidered`	`number`	Total candidate memories evaluated
`tokenCount`	`number`	Estimated token count of the context paragraph

If no relevant memories are found, context will be empty.

curl Example

curl -X POST https://mem.anishroy.com/api/v1/memory/prepare \
  -H "Authorization: Bearer sk_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "namespace": "my-app",
    "entity": "user-123",
    "messages": [
      { "role": "user", "content": "Help me plan dinner tonight" }
    ]
  }'

LLM Cost

2 LLM calls per request — one for intent analysis, one for context synthesis. See prepare() SDK reference for pipeline details.