Persistent Task Planning (Manus-style)
Manages complex AI coding tasks through persistent, markdown-based planning and state tracking on disk.
Manages complex AI coding tasks through persistent, markdown-based planning and state tracking on disk.
This skill implements a 'Manus-style' persistent planning framework that treats the AI's context window as volatile RAM and the local filesystem as stable disk storage. By maintaining structured files like task_plan.md, findings.md, and progress.md within a .kiro/plan/ directory, it ensures that long-running agentic tasks remain crash-proof and survive context clearing. This approach provides a deterministic completion gate and shared state, making it ideal for complex refactors, deep research, and multi-step workflows where context drift would otherwise lead to failure.
