https://malaev.dev/en/tags/agents/Recent content in Agents on malaev.devHugoen-USWed, 09 Sep 2026 00:00:00 +0000https://malaev.dev/en/posts/compaction-is-cheaper-than-the-context/Wed, 09 Sep 2026 00:00:00 +0000https://malaev.dev/en/posts/compaction-is-cheaper-than-the-context/<p>When a coding agent&rsquo;s context fills up, it gets compacted: the history is replaced by a summary and the session continues. Everyone who has watched this happen mid-task has the same instinct — compact as late as possible. The summary loses detail, and compacting throws away the prompt cache you have been paying to build. Better to hold on.</p> <p>Half of that is true. The other half is expensive, and I had the threshold set wrong for weeks because of it.</p>https://malaev.dev/en/posts/agent-memory-in-the-repo/Tue, 08 Sep 2026 00:00:00 +0000https://malaev.dev/en/posts/agent-memory-in-the-repo/<p>A coding agent starts every session knowing nothing. You explain the project, it does good work, the session ends, and the next one begins from zero. The usual patch is a longer context window, but that only moves the wall. When the context fills, it gets summarised — and a summary keeps decisions while losing the reasons behind them. &ldquo;We chose A&rdquo; survives. &ldquo;We rejected B because we measured it and it was worse&rdquo; does not. So the next session proposes B again, and you pay for that experiment twice.</p>