THREAT OPS › Threat News › Context Engineering | Compaction & Agent Memory for Automated Malware Analysis
Context Engineering | Compaction & Agent Memory for Automated Malware Analysis
<h2>Executive Summary</h2> <ul> <li>Compaction is a context-management pattern used across agent systems to compress prior context into a denser working state for long-running tasks.</li> <li>SentinelLABS evaluated OpenAI’s native Responses API implementation against our automated malware analysis evaluation harness to measure real-world impact on task quality and cost.</li> <li>Compaction reduced
MITRE ATT&CK techniques
- CompressionT1027.015
Indicators of compromise
- https://openai.com/index/equip-responses-api-computer-environmenturl
- https://www.anthropic.com/engineering/effective-harnesses-for-long-running-agentsurl
- https://docs.langchain.com/oss/python/deepagents/context-engineeringurl
- https://jxnl.co/writing/2025/08/30/context-engineering-compactionurl
- https://arxiv.org/abs/2601.07190url