planning-with-files

You can maintain task planning state by using persistent on-disk memory files like task_plan.md, findings.md, and progress.md to capture plans, discoveries, and phase tracking across multiple tool calls. This prevents losing your goals and context during long-running research tasks.

Session recovery after a clear is handled automatically through on-disk working memory. The Skill uses recovery scripts and persistent markdown templates to restore prior context, plans, and progress so you can resume your multi-step tasks seamlessly.

Yes, you can use markdown templates to track progress for multi-phase tasks. The Skill enforces the creation of task_plan.md, findings.md, and progress.md, allowing you to log discoveries and update your task plan before beginning each new phase.

To create a planning skeleton, start by defining phases in a task_plan.md file within your project, alongside findings.md and progress.md. You then update this plan before each phase to log progress and maintain a structured workflow throughout execution.

Complex task planning loses goals because intermediate state is not persisted. You prevent this by maintaining a persistent on-disk working memory of planning artifacts, ensuring your plans, findings, and progress are safely captured across all tool calls.

No dependencies are required to manage workflows with on-disk working memory. The Skill operates independently using scripts to create and maintain task_plan.md, findings.md, and progress.md files for your project planning and session recovery.