moai-cc-memory

Context window overflow in extended sessions occurs when accumulated code, logs, and tool outputs exceed token limits. This Skill manages session memory through just-in-time retrieval, layered context summaries, and memory-file patterns to keep your working context within 100K–200K token windows while preserving access to prior work.

Memory files are structured entries that capture ongoing task state, decisions, and file references for quick retrieval. Create a memory entry detailing your feature progress, then reference it in later steps to hand off context efficiently between sessions without reloading entire projects.

Yes. Selective file loading loads only relevant files on demand rather than the entire project, while result caching stores computed outputs. Together with memory-file management, these techniques enable efficient workflow across large projects by controlling what stays in active context.

Just-in-time retrieval fetches files and cached results only when needed, while memory management organizes session state into structured memory files. This combination prevents context bloat by deferring non-critical data retrieval until the task requires it.

Yes. This Skill is built to integrate with Read, Write, Glob, and Bash tools to enable efficient context management. These tools support selective file loading, memory-file operations, and result caching across your workflow.

Apply memory management for long-running sessions, large projects, and cross-team handoffs where context budgeting is critical. It's essential when token usage approaches limits and you need reliable selective loading and memory-file patterns to maintain efficiency.