Day 1 · Block 6

E06 - Memory behavior with session and long-term store

• Why would your agent need memory?
• How is memory different from context? How does it relate to tools?
• Take a look at OpenClaw. What is it? How is it different from your coding agent? What are the benefits and risks of using it? https://docs.openclaw.ai/
• Run `python exercises/06/memory_demo.py`.
• Try one `TODO-STUDENT` modification and rerun.
• Show short-term reset behavior and long-term retrieval behavior.

Inputs

• exercises/06/memory_demo.py
• Session reset behavior
• long_term_store.json

Deliverable

memory trace proving short-term vs long-term distinction + one risk note.

Target

Why would your agent need memory?

Submission

• Use the live launcher: Live Exercises
• Direct prefilled form: E06 submission link

Checklist

• Make one explicit design decision.
• Include one verification check.
• State one limitation or risk.

Common failure modes

• Memory behavior claimed without trace evidence.
• Unsafe or stale memory retained without correction.

Extension task (optional)

Compress memory notes and re-evaluate retrieval quality.