planning-with-files

Maintain task planning across multiple sessions by writing goals, findings, and status updates to persistent markdown files like task_plan.md, findings.md, and progress.md stored on disk. This allows you to track long-running workflows and automatically recover session context when resuming work.

The best way to track findings and phase status for complex research projects is using on-disk markdown files. You record discoveries in findings.md and update phase status in progress.md, ensuring auditable execution and alignment with end goals during multi-tool workflows.

Initialize planning files for a multi-tool workflow by running the provided init-session.sh script in your project directory. This sets up the required task_plan.md, findings.md, and progress.md artifacts to begin tracking your task phases, discoveries, and overall progress.

Yes, you can verify task completion automatically by running the check-complete.sh script. This script checks your planning markdown files to ensure all phases and goals defined in task_plan.md have been successfully addressed before finalizing the work.

File-based planning works effectively for recovering context after a session interruption by utilizing automatic session recovery and planning refresh mechanisms. Scripts like session-catchup.py read your progress.md and task_plan.md to reload previous findings and goals seamlessly.

Use on-disk markdown planning instead of session memory for long-running tasks, research projects, or multi-tool workflows where phases take significant time. It ensures safe, auditable execution and keeps goals in focus by reading the plan before making major decisions.