planning-with-files

Persist task plans, findings, and progress to markdown files.

Updated Jul 11, 2022
One-click install
npx skills add https://github.com/zhengfran/dotconfig --skill planning-with-files-zhengfran
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: planning-with-files
Source: https://github.com/zhengfran/dotconfig/tree/main/tools/ai/agents/skills/planning-with-files
Command: npx skills add https://github.com/zhengfran/dotconfig --skill planning-with-files-zhengfran

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Prevents context loss during long, multi-step tasks by using persistent markdown files as working memory so goals, decisions, findings, and progress remain available across many tool calls and sessions.

Core Features & Use Cases

  • Disk-based working memory: Creates and maintains task_plan.md, findings.md, and progress.md in your project root to keep plans and knowledge durable.
  • Manus-style planning workflow: Enforces a phase-based approach with decision logs, error tracking, and “read-before-decide” discipline.
  • Session recovery: Supports automatic detection of unsynced context from previous sessions and guides you to reconcile it before continuing.
  • Security guardrails for untrusted content: Reduces prompt-injection risk by directing external/browser/search results to findings.md rather than task_plan.md.

Quick Start

Run a long multi-step plan by creating task_plan.md, findings.md, and progress.md in your project root, then update task_plan.md after each phase and write new discoveries into findings.md.

Frequently Asked Questions about planning-with-files

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prevent context loss during complex, multi-step tasks in an agent session?

To prevent context loss during multi-step tasks, persistent disk memory maintains your plans, findings, and progress in markdown files. This file-based working memory ensures goals remain available across many tool calls and sessions.

How do I recover an agent session after losing context during feature development?

Session recovery works by automatically detecting unsynced context from previous sessions stored in markdown files. It guides you to reconcile findings and progress data before continuing your feature development work.

What is file-based memory for task management and when do I need it?

File-based memory for task management uses markdown files like task_plan.md and findings.md as durable working memory. You need it for research or implementation tasks requiring many tool calls and reliable resume behavior.

How do I defend against prompt injection when using external content in task planning?

To defend against prompt injection from external content, direct browser and search results to findings.md rather than task_plan.md. This security guardrail isolates untrusted external data from your core planning files.

How do I start a phase-based planning workflow using markdown files?

Start a phase-based planning workflow by creating task_plan.md, findings.md, and progress.md in your project root. Update task_plan.md after each phase and write new discoveries into findings.md as you work.

Does file-based planning work for long research tasks that span multiple sessions?

File-based planning works for long research tasks by persisting plans, findings, and progress to markdown files on disk. This approach reliably supports session recovery and maintains context across multiple work sessions.