Memory as Infrastructure: Reliability Engineering for Persistent Agent Memory in Months-Long LLM-Assisted Development
Source
Mike Helwig
Author
Mike Helwig
Date
Terms in this piece · Glossary
agent harness — The scaffolding around a model that turns it into a working agent — the loop, the tools it can call, and the rules for when to stop.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
context compaction — Summarizing an agent's earlier conversation to free room in the context window so a long session can keep going.
Why it matters
This account of a months-long continuous Claude Code session describes SIx agent harnessThe scaffolding around a model that turns it into a working agent — the loop, the tools it can call, and the rules for when to stop.Full definition →, an open-source AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → memory layer with session-start health gates and anti-recurrence stores so memory survives repeated context compactionSummarizing an agent's earlier conversation to free room in the context window so a long session can keep going.Full definition →.
articleWhen Does Memory Help? A Cost-Aware Evaluation of Long-Term Memory in Tool-Using LLM AgentsShweta Mishra, Shashank Mishra
articleFilesystem-Based Memory for LLM Agents: Organization, Evolution, and SustainabilitySizhe Zhou, Sheldon Yu, Hui Wei, Junda Wu, Siru Ouyang, Yizhu Jiao, Shijia Pan, Julian McAuley, Yu Zhang, Tong Yu, Jiawei Han
articleCoupling Planning with Episodic Memory in LLM Agents for Software Issue ResolutionJiahao Zhang, Yifan Zhang, Yu Huang