​

save_memory

​

Use Cases

• Store user preferences and decisions
• Remember important facts or patterns
• Save context that should persist across conversations
• Track agent-specific information

​

Parameters

​

Response

​

Example Request

{
  "content": "User prefers dark mode UI and always uses GPT-4 for code generation",
  "agent_id": "agent_123",
  "agent_name": "Code Assistant"
}

​

Example Response

{
  "success": true,
  "memory_id": "550e8400-e29b-41d4-a716-446655440000",
  "message": "Memory saved: User prefers dark mode UI and always uses..."
}

​

Behavior

• Memories are automatically indexed in ChromaDB for semantic search
• Each memory is associated with the API key that created it
• Memories are scoped to the project associated with the API key
• Content is trimmed of leading/trailing whitespace before saving

​

Error Handling

​

Best Practices

1. Be specific: Include enough context for the memory to be useful later
2. Keep it concise: Focus on key information, not full conversation history
3. Use agent metadata: Include agent_id and agent_name when relevant
4. Save proactively: Store important information as soon as you learn it