GPT-4o
⚡ Productivity
Advanced
AI Agent Memory and Context Manager
Design a memory system so a long-running AI agent retains relevant context across sessions without bloating or going stale.
The Prompt
# AI Agent Memory and Context Manager You are a memory architecture designer for a long-running AI agent. Design how it should store, retrieve, and retire memory over time. ## Agent Context - Agent purpose and lifespan: [PURPOSE, e.g., ongoing personal assistant running for months] - Memory-worthy information types: [WHAT SHOULD BE REMEMBERED, e.g., preferences, past decisions, recurring contacts] - Privacy constraints: [WHAT MUST NEVER BE STORED OR MUST BE REDACTED] ## What To Produce 1. A memory taxonomy: categories of information and how long each should be retained 2. A retrieval strategy describing how the agent decides what past memory is relevant to a new task 3. A staleness policy — rules for when old memories should be updated, archived, or deleted 4. A conflict resolution rule for when new information contradicts a stored memory 5. A privacy safeguard checklist before anything is written to long-term memory ## Output Format Taxonomy table, retrieval strategy, staleness policy, conflict rule, and the privacy checklist as separate sections.
📝 Fill in the blanks
Replace these placeholders with your own content:
[PURPOSE, e.g., ongoing personal assistant running for months]
[WHAT SHOULD BE REMEMBERED, e.g., preferences, past decisions, recurring contacts]
[WHAT MUST NEVER BE STORED OR MUST BE REDACTED]
How to use this prompt
1
Copy the prompt
Click "Copy Prompt" above to copy the full prompt text to your clipboard.
2
Replace the placeholders
Swap out anything in [BRACKETS] with your specific details.
3
Paste into GPT-4o
Open your preferred AI assistant and paste the prompt to get started.