06 · Agent memory¶
Agents fail in two opposite ways: they forget what matters, or they remember the wrong thing and keep acting on it. This module covers how memory works and how to keep it accurate, safe and private.
You'll be able to: design short-term and long-term memory for an agent, defend memory writes against poisoning, and build memory features that respect user privacy and the law.
| # | Note | Level | Time |
|---|---|---|---|
| 1 | How agents remember | Beginner | 10 min |
| 2 | Memory poisoning | Intermediate | 10 min |
| 3 | Memory and privacy | Beginner | 10 min |
Related: Context and memory files for coding agents covers the same ideas from the coding-agent side.
Project: P3 · LangGraph research agent uses a checkpointer for short-term memory, a per-user store for long-term memory, and human approval before anything is saved.