Memory Configuration Memory

Configure the long-term memory system.

Overview

The memory system allows AI to remember user preferences and past interactions.

Basic Configuration

memory:
  enabled: true           # Enable memory system
  maxMemories: 1000       # Max memories per user
  autoExtract: true       # Auto-extract from conversations
  similarityThreshold: 0.7  # Retrieval threshold

Configuration Options

Option	Type	Default	Description
enabled	boolean	true	Enable memory
maxMemories	number	1000	Max per user
autoExtract	boolean	true	Auto-extract memories
similarityThreshold	number	0.7	Match threshold
retrievalCount	number	5	Max memories per query

Memory Types

Type	Description	Example
Preference	User preferences	"Likes jazz music"
Fact	Personal facts	"Birthday is May 15"
Context	Conversation context	"Working on a Python project"

Manual Memory Commands

#记忆 添加 喜欢周杰伦的歌    # Add memory
#记忆 列表                   # List memories
#记忆 删除 1                 # Delete by ID
#清除记忆                    # Clear all memories

API Access

// Add memory
await fetch('/api/memory', {
  method: 'POST',
  body: JSON.stringify({
    userId: '123456789',
    content: 'Prefers dark mode'
  })
})

// Get memories
const memories = await fetch('/api/memory?userId=123456789')

Storage Backend

Default: SQLite with vector extension

memory:
  storage:
    type: sqlite           # sqlite, postgres
    path: ./data/memory.db

Privacy

User Privacy

• Memories are per-user and private
• Users can view and delete their memories
• Consider data retention policies

memory:
  retention:
    maxAge: 365            # Days to keep memories
    autoCleanup: true      # Auto cleanup old memories

Next Steps

• Context - Context management
• Features - Feature settings