Skip to content

Agent Memory

Give agents personal knowledge base to learn and improve over time.

What is Agent Memory

Agent Memory enables agents to store and recall their own knowledge separately from shared user preferences.

Memory Type Owner Access Purpose
User Memory You All agents Your preferences and requirements
Agent Memory Each agent Only that agent Agent's learned patterns and experiences

How Agents Use Memory

Agents manage memory autonomously during conversations:

Before tasks: Search memories for relevant patterns

During execution: Add learnings and update knowledge

After completion: Store successful approaches

Key Benefits

Benefit Result
Persistent learning Knowledge accumulates across conversations
Personalized service Agents remember your style and preferences
Autonomous improvement Agents learn from experience without manual input
Context isolation Each agent builds domain-specific expertise

Managing Memory

Via Memory Manager

Open Memory Manager from workspace to view and manage any agent's knowledge:

  • View categories and items
  • Add, edit, or delete memories
  • Export as JSON, Markdown, TXT, or YAML
  • Import content with AI parsing

Via Memory Tools

Agents use built-in memory tools during conversations:

  • list_memory_categories - Discover existing knowledge
  • get_memory - Retrieve specific memories
  • add_memory - Store new learnings
  • update_memory - Refine existing knowledge
  • search_memory_with_agent - AI-powered search in isolated context

Common Use Cases

Agent Type Memory Strategy Outcome
Code Reviewer Stores style rules in code_standards Consistent reviews
Support Agent Saves solutions in faq_answers Faster responses
Writing Agent Remembers tone in brand_voice Unified style