The article explains why default agent memory is fleeting, proposes a two‑layer design of session and long‑term memory with a post‑session “dreaming” integration step, and shows how selective persistence and shared long‑term storage keep agents continuously improving.
The article systematically breaks down the six essential building blocks of a programming agent—live repository context, prompt shape and cache reuse, structured tool access and validation, context reduction, structured session memory, and bounded sub‑agent delegation—illustrated with a Mini Coding Agent implementation and comparisons to Claude Code, Codex, and OpenClaw.
The article dissects Claude Code's three‑level progressive compression system—MicroCompact, SessionMemoryCompact, and Full Compact—showing how it edits cached prompts, maintains background memory files, and generates a structured nine‑section summary to keep AI agents operating over arbitrarily long conversations within a limited context window.
Claude Code forgets all prior context each new session because large language models only see the current window, but the open‑source claude‑mem plugin records project actions, compresses them into semantic summaries, and injects the relevant history back into Claude, dramatically reducing re‑explanation overhead.
