Stop using the model as your memory
The author describes how large language models can lose track of decisions and revert to outdated information during extended interactions. They propose storing the project state in external specifications and checklists rather than relying on the model's context window as memory. This approach reduces token waste and improves consistency for both personal and production agents.
- ▪The model repeatedly forgot earlier decisions even when using a 1 million token context window, leading to repeated rework.
- ▪The author moved the state to a frozen specification file and a verifiable checklist, treating the model as a worker rather than a filing cabinet.
- ▪Session measurements showed that most tokens were reused instructions rather than fresh input, and shrinking the per‑step context improved performance.
- ▪Production agents often fail because they cannot reload the full conversation, whereas external state files allow them to resume reliably.
Opening excerpt (first ~120 words) tap to expand
try { if(localStorage) { let currentUser = localStorage.getItem('current_user'); if (currentUser) { currentUser = JSON.parse(currentUser); if (currentUser.id === 3937147) { document.getElementById('article-show-container').classList.add('current-user-is-article-author'); } } } } catch (e) { console.error(e); } greymoth Posted on Jun 26 Stop using the model as your memory #ai #claudecode #agents #productivity I run Claude Code most of the day. The thing that kept biting me wasn't the model getting dumber. It was the model forgetting what we'd already settled, then confidently redoing it wrong. You've probably hit it. You write a CLAUDE.md, you keep notes, you tell it "we decided X." A few prompts later it relitigates X, or quietly breaks something it fixed an hour ago.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at DEV.to (Top).
Discussion
0 commentsMore from DEV.to (Top)
