This skill implements a robust 'working memory on disk' approach inspired by the Manus workflow, designed specifically for complex and multi-step tasks. It mandates a three-file architecture—task_plan.md, notes.md, and a final deliverable—to ensure that project goals, research findings, and decision logs remain persistent across long-running sessions. By externalizing information to the file system, this skill prevents context window bloat, reduces hallucination in complex projects, and provides a clear audit trail for error recovery and status updates.
Key Features
01Context optimization by offloading data to the file system
02Persistent task state management via task_plan.md
03Structured research capture with notes.md templates
04Manus-style three-file architectural pattern
05Comprehensive error logging and decision tracking
060 GitHub stars
