Memory, State & Execution

Distinguish the three memory layers every agent uses — short-term working context, long-term learned facts, and external retrieval — and pick the right layer for each piece of state so the agent doesn't blow the context window or leak stale data.

Every memory record needs a scope — user, repo, session, or task. Get the scope wrong and the agent leaks one user's preferences into another's session, or worse, applies last week's plan to today's PR. This topic teaches you to bind memory to the smallest scope that still makes it useful.

Memory that never expires becomes a liability. This topic covers TTLs, recency- and relevance-based pruning, and explicit reset rules so the agent's context stays compact, correct, and forgettable when the user asks.

Persisting agent state is half the job — the other half is noticing when that persisted state has drifted from reality. This topic covers durable state stores, version stamps, and the failure signals that tell you the agent is acting on a stale snapshot.

A real agent run can outlive a single process: pods restart, users go to bed, sessions get handed off. This topic covers the durable artifacts and checkpoints that let an agent resume mid-task instead of starting over.

Agents rarely live in one tool. A run may start in an IDE, continue in a chat, hand off to a CI bot, and finish in a code-review UI. This topic covers how to carry identity, context, and state across those boundaries without losing the thread.