othmanadi/planning-with-files

Overview

This skill implements Manus-style file-based planning to manage complex, multi-step tasks using persistent markdown files. It automatically creates and updates task_plan.md, findings.md, and progress.md and includes automatic session recovery for unsynced work. Use it to keep decisions, discoveries, and session logs on disk so nothing important is lost between tool calls.

How this skill works

The skill initializes three markdown files in your project root and enforces a write-first workflow: save key findings after every two external lookups and read the plan before major decisions. It logs phases, errors, and session progress and offers a session-catchup script to recover unsynced context after interruptions. Helper scripts automate initialization and completion checks.

When to use it

• Starting any complex task with 3+ steps
• Research projects that require repeated lookups
• Workflows that call tools or browsers more than five times
• Long-running tasks you may pause and resume
• Projects where reproducible error logging matters

Best practices

• Create task_plan.md before you start executing any steps
• Save findings immediately after every two view/search operations
• Read task_plan.md and findings.md before making major decisions
• Log all errors and update phase status after each phase
• Keep planning files in the project root, not the skill installation folder

Example use cases

• Multi-phase feature development with iterative testing and debugging
• Literature or competitive research spanning many browser lookups
• Migrating or refactoring codebases where steps and failures must be tracked
• Data analysis pipelines that require repeated tool runs and checkpointing
• Long-term experiments where session recovery and reproducibility are essential