Checkpoint: Memory Mastery
Checkpoint: Memory Mastery
You've completed the core module. Let's verify your understanding.
Knowledge Check
Answer these questions (to yourself or write them down):
-
What's the difference between short-term and long-term memory for an agent?
-
Why is MEMORY.md an index rather than a full knowledge base?
-
What should happen before context compaction?
-
Why is forgetting considered a feature?
-
What are the memory layers in the Cortex Model?
Practical Check
Your system should now have:
Files
- MEMORY.md (under 3KB, with pointers)
- memory/WORKING.md
- memory/daily/[today].md with real entries
- At least one detail file (person or project)
- USER.md with your information
Behavior
- Agent loads core files at session start
- Agent can find facts you've stored
- Agent knows to save before compaction
- Agent uses appropriate memory tags ([E], [L], [K])
Understanding
- You can explain the Cortex Model
- You know when to load vs when to search
- You understand why compression matters
- You have a plan for memory maintenance
The Test
Start a new session with your agent and ask:
"What did we work on yesterday?"
If your agent can answer accurately from memory, you've mastered this module.
If not, review the material and fix your memory structure.
This is the most important module. If memory works, everything else gets easier. If it doesn't, nothing else matters.
Take your time here. Get it right.