概要
Filesystem Context Engineering provides a sophisticated framework for managing information within AI agent systems, specifically addressing the limitations of fixed context windows. By treating the filesystem as a persistent memory layer, this skill allows agents to offload massive tool outputs, maintain plan persistence across long-duration tasks, and facilitate high-fidelity communication between sub-agents. It transitions workflows from static context loading to a dynamic discovery model, utilizing targeted search tools to pull only relevant data into the prompt, thereby significantly reducing token bloat and improving reasoning accuracy.
