planning-with-files

Organizes complex tasks into persistent markdown planning files for progress tracking and session recovery.

Updated Jan 31, 2026

Search Tags:#workflow #planning #markdown #context management #task tracking #session recovery #agent memory

Ready to run

npx skills add https://github.com/adrianasiam19/ATLAS --skill planning-with-files-adrianasiam19

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: planning-with-files
Source: https://github.com/adrianasiam19/ATLAS/tree/main/.agent/skills/planning-with-files
Command: npx skills add https://github.com/adrianasiam19/ATLAS --skill planning-with-files-adrianasiam19

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve? Long multi-step tasks cause AI agents to lose track of goals after many tool calls, and context resets erase progress. This Skill writes plans, findings, and progress logs to disk so work survives context limits and session interruptions. ## Core Features & Use Cases - File-Based Planning: Creates task_plan.md, findings.md, and progress.md in your project directory to track phases, decisions, and errors. - Session Recovery: Detects unsynced context from previous sessions via a catchup script and supports automatic recovery after /clear. - Structured Templates: Ships generic and analytics-specific templates plus init and completion-check scripts for bash, PowerShell, and Python. - Use Case: When asked to research a topic or build a multi-phase feature, the Skill creates a phase plan first, logs findings after every search, and resumes cleanly even after the conversation is cleared. ## Quick Start Ask the AI to plan out and organize your multi-step project using file-based planning before starting any work.

Frequently Asked Questions about planning-with-files

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

FAQPage Schema

How do I keep an AI agent on track during long multi-step tasks?▼

Write the plan to a persistent file like task_plan.md and re-read it before major decisions. This keeps goals in the model's recent attention window even after dozens of tool calls, preventing the lost-in-the-middle effect.

How does file-based planning recover after clearing the conversation?▼

The Skill stores task_plan.md, findings.md, and progress.md in your project directory. On a new session, a catchup script scans the previous session transcript for unsynced context, then you run git diff and update the planning files before continuing.

When should I use file-based planning instead of a todo list?▼

Use it for tasks with three or more steps, research work, or anything spanning many tool calls. Skip it for simple questions, single-file edits, or quick lookups where persistent state adds no value.

Does planning-with-files work on Windows and with Codex?▼

Yes. It ships PowerShell scripts (init-session.ps1, check-complete.ps1) alongside bash versions, and the session-catchup.py script parses both Claude Code and Codex session transcripts to find unsynced context.

Why should web search results not go into task_plan.md?▼

A PreToolUse hook re-reads task_plan.md before every tool call, so untrusted web content written there would be repeatedly injected into context, amplifying prompt injection risk. External content belongs in findings.md only.