planning-with-files

Persistent planning uses markdown files like task_plan.md, findings.md, and progress.md to store task phases and logs, preventing information loss across context windows and enabling automatic session recovery.

File-based planning leverages markdown files as external memory to manage multi-step tasks. It structures task phases, research discoveries, and progress logs to ensure meticulous tracking and task completion.

Automatic session recovery checks for and reconciles unsynced context from previous sessions. It reads persistent markdown files to restore your task phase, findings, and progress logs seamlessly.

Yes, the approach utilizes shell scripts to initialize sessions and markdown files for persistent memory. It supports standard tools including Read, Write, Edit, Bash, Glob, Grep, WebFetch, and WebSearch.

Error handling implements a 3-strike protocol for recurring issues and encourages logging all encountered problems directly into the persistent markdown files for later review and resolution.

You need persistent markdown files when working on large research projects or developing new features where context window limitations risk losing critical details, phases, and research findings.