planning-with-files

To maintain planning memory for multi-phase tasks, you can use file-based planning to create task_plan.md, findings.md, and progress.md. These files act as persistent memory to keep goals and discoveries aligned across long sessions.

The best way to track research progress across many tool calls is by logging discoveries in a dedicated findings.md file and updating progress.md. Re-reading these files ensures your research stays aligned with the original task plan.

To initialize planning files for a new complex project, run the setup script with your project name. This bootstrap process uses predefined templates to automatically generate task_plan.md, findings.md, and progress.md in your working directory.

Yes, file-based planning works without external dependencies because it relies entirely on local markdown files for persistent memory. It uses built-in templates and optional scripts to automate setup without requiring external libraries.

You should use persistent markdown files for task planning when working on multi-phase tasks or research projects that span many tool calls. This approach prevents context loss and keeps your planning aligned during extended work sessions.

Persistent working memory in file-based task planning is a method of storing goals, discoveries, and session logs in markdown files. It ensures that complex task planning remains accessible and aligned without relying on volatile session context.