starred/planning-with-files

Fork 0

mirror of https://github.com/OthmanAdi/planning-with-files.git synced 2026-04-26 08:26:01 +03:00

[GH-ISSUE #17] Understanding Planning with files #10

New issue

Closed

opened 2026-03-03 18:49:56 +03:00 by kerem · 1 comment

kerem commented

2026-03-03 18:49:56 +03:00

Owner

Originally created by @SchneiderSam on GitHub (Jan 11, 2026).
Original GitHub issue: https://github.com/OthmanAdi/planning-with-files/issues/17

Hi,

sorry for opening an "issue" for an question but culdn't find any discord channel or discussion forum. So my question is: whats the difference between this repo, Taskmaster, bMad Method, Open Spec ect? Isn't it the same?

Originally created by @SchneiderSam on GitHub (Jan 11, 2026). Original GitHub issue: https://github.com/OthmanAdi/planning-with-files/issues/17 Hi, sorry for opening an "issue" for an question but culdn't find any discord channel or discussion forum. So my question is: whats the difference between this repo, Taskmaster, bMad Method, Open Spec ect? Isn't it the same?

kerem closed this issue

2026-03-03 18:49:56 +03:00

kerem commented

2026-03-03 18:49:57 +03:00

Author

Owner

@OthmanAdi commented on GitHub (Jan 11, 2026):

Hallo Samuel!

Es ist aufjeden falls n toller frage!! Let me break down the actual differences based on what these methods really are.

The Short Answer

They're similar in using files for AI agent planning, but very different in approach and implementation. Here's what each actually does:

The Real Differences

Planning-with-Files (This Repo)

What it is:

Based on Manus AI's architecture (the $2B Meta acquisition)
Implements "context engineering" through persistent markdown files
Built specifically for Claude Code v2.1+ with automatic hooks

The Pattern:

3 files: task_plan.md (phases/progress), findings.md (research), progress.md (session log)
Hooks automate everything: PreToolUse reads the plan before decisions, PostToolUse reminds you to update status, Stop blocks completion until done
Philosophy: "Filesystem = persistent memory" — prevents AI from forgetting goals after many tool calls

Use when: You want Claude to automatically manage its own planning files without you having to tell it

TaskMaster AI

What it is:

An AI-powered task management system you drop into Cursor, Lovable, Windsurf, Roo, etc.
Works as a "project manager" for your AI agent
Uses a .taskmaster/ folder structure

The Pattern:

PRD-driven: You create .taskmaster/docs/prd.txt (Product Requirements Document)
Task decomposition: Automatically breaks PRD into tasks and subtasks
Multi-role AI: Supports different AI providers (OpenAI, Anthropic, Gemini) with main/research/fallback roles
Task files: Generates individual task markdown files with status tracking

Use when: You have a PRD and want AI to automatically generate and manage tasks from it

Sources:

BMAD Method

What it is:

"Breakthrough Method for Agile AI-Driven Development"
A complete framework for AI-driven development with specialized planning agents
YAML-based workflow orchestration

The Pattern:

Agentic Planning: Dedicated AI agents (Analyst, PM, Architect) collaborate with you upfront
Creates comprehensive specs: PRDs and Architecture documents through iterative refinement
YAML workflows: Structured blueprints that orchestrate tasks across different agents
Human-in-the-loop: Advanced prompt engineering with your input to create robust plans
Scale-adaptive: Planning adjusts based on project size

Use when: You want specialized AI agents to help you create detailed specifications before building

Sources:

Open Agent Specification (Oracle)

What it is:

A standardized way to define AI agents (not a planning method)
Framework-agnostic declarative language for agent portability
Makes agents work across different AI frameworks

The Pattern:

Agent definitions: Describes agent behavior, tools, and workflows in portable config
Interoperability: Define once, run on any compatible framework
Not really about planning: More about agent architecture and portability

Different use case: This is about making agents portable, not about task planning

Sources:

Spec-Driven Development (GitHub)

What it is:

An approach where you write specifications first, then AI codes from them
Part of GitHub's spec-kit toolkit
Creates specs → plans → tasks → implementation

The Pattern:

Write detailed spec docs upfront
AI generates implementation plan
Breaks into small, reviewable chunks
Each task implemented and tested in isolation

Use when: You want to drive development from written specifications

Sources:

Quick Comparison Table

Method	Focus	Best For	File Structure
planning-with-files	Context engineering for Claude	Preventing AI goal drift during execution	3 markdown files with hooks
TaskMaster AI	PRD-based task management	Breaking down requirements into tasks	`.taskmaster/` folder with PRD + tasks
BMAD Method	Agentic planning with specialized agents	Creating comprehensive specs upfront	YAML workflows + PRD + Architecture docs
Open Agent Spec	Agent portability	Making agents work across frameworks	Agent definition configs
Spec-Driven Dev	Specification-first coding	Driving implementation from detailed specs	Spec docs → plans → tasks

Why This Repo is Different

After Manus sold for $2B, people realized their secret was forcing AI to re-read goals constantly to prevent context drift.

This repo is the only one that:

Uses Claude Code v2.1+ hooks to automate the pattern
Makes the AI read task_plan.md before every major decision (PreToolUse hook)
Blocks completion until all phases are done (Stop hook)

The others require you to manage the process. This one makes Claude manage it automatically.

Try It

/plugin marketplace add OthmanAdi/planning-with-files
/plugin install planning-with-files@planning-with-files

Start a complex task and Claude will automatically handle the planning files without you telling it to.

Hope this clears it up! The methods aren't really competing, they solve different problems. You could even use BMAD for upfront planning, then planning-with-files for execution tracking.

And yeah, we should add Discussions! Good call.

— Ahmad

@OthmanAdi commented on GitHub (Jan 11, 2026): Hallo Samuel! Es ist aufjeden falls n toller frage!! Let me break down the actual differences based on what these methods really are. --- ## The Short Answer They're similar in **using files for AI agent planning**, but very different in **approach and implementation**. Here's what each actually does: --- ## The Real Differences ### **Planning-with-Files** (This Repo) **What it is:** - Based on [Manus AI's architecture](https://manus.im/blog/Context-Engineering-for-AI-Agents-Lessons-from-Building-Manus) (the $2B Meta acquisition) - Implements "context engineering" through persistent markdown files - Built specifically for Claude Code v2.1+ with automatic hooks **The Pattern:** - **3 files:** `task_plan.md` (phases/progress), `findings.md` (research), `progress.md` (session log) - **Hooks automate everything:** PreToolUse reads the plan before decisions, PostToolUse reminds you to update status, Stop blocks completion until done - **Philosophy:** "Filesystem = persistent memory" — prevents AI from forgetting goals after many tool calls **Use when:** You want Claude to automatically manage its own planning files without you having to tell it --- ### **[TaskMaster AI](https://github.com/eyaltoledano/claude-task-master)** **What it is:** - An AI-powered task management system you drop into Cursor, Lovable, Windsurf, Roo, etc. - Works as a "project manager" for your AI agent - Uses a `.taskmaster/` folder structure **The Pattern:** - **PRD-driven:** You create `.taskmaster/docs/prd.txt` (Product Requirements Document) - **Task decomposition:** Automatically breaks PRD into tasks and subtasks - **Multi-role AI:** Supports different AI providers (OpenAI, Anthropic, Gemini) with main/research/fallback roles - **Task files:** Generates individual task markdown files with status tracking **Use when:** You have a PRD and want AI to automatically generate and manage tasks from it **Sources:** - [TaskMaster GitHub](https://github.com/eyaltoledano/claude-task-master) - [Tutorial](https://github.com/eyaltoledano/claude-task-master/blob/main/docs/tutorial.md) - [Shipixen guide](https://shipixen.com/tutorials/reduce-ai-coding-errors-with-taskmaster-ai) --- ### **[BMAD Method](https://github.com/bmad-code-org/BMAD-METHOD)** **What it is:** - "Breakthrough Method for Agile AI-Driven Development" - A complete framework for AI-driven development with specialized planning agents - YAML-based workflow orchestration **The Pattern:** - **Agentic Planning:** Dedicated AI agents (Analyst, PM, Architect) collaborate with you upfront - **Creates comprehensive specs:** PRDs and Architecture documents through iterative refinement - **YAML workflows:** Structured blueprints that orchestrate tasks across different agents - **Human-in-the-loop:** Advanced prompt engineering with your input to create robust plans - **Scale-adaptive:** Planning adjusts based on project size **Use when:** You want specialized AI agents to help you create detailed specifications before building **Sources:** - [BMAD GitHub](https://github.com/bmad-code-org/BMAD-METHOD) - [What is BMAD Method?](https://medium.com/@visrow/what-is-bmad-method-a-simple-guide-to-the-future-of-ai-driven-development-412274f91419) - [Mastering BMAD](https://medium.com/@courtlinholt/mastering-the-bmad-method-a-revolutionary-approach-to-agile-ai-driven-development-for-modern-e7be588b8d94) --- ### **[Open Agent Specification](https://arxiv.org/html/2510.04173v1)** (Oracle) **What it is:** - A standardized way to **define AI agents** (not a planning method) - Framework-agnostic declarative language for agent portability - Makes agents work across different AI frameworks **The Pattern:** - **Agent definitions:** Describes agent behavior, tools, and workflows in portable config - **Interoperability:** Define once, run on any compatible framework - **Not really about planning:** More about agent architecture and portability **Different use case:** This is about making agents portable, not about task planning **Sources:** - [Oracle announcement](https://blogs.oracle.com/ai-and-datascience/introducing-open-agent-specification) - [Technical report](https://arxiv.org/html/2510.04173v1) --- ### **[Spec-Driven Development](https://github.com/github/spec-kit)** (GitHub) **What it is:** - An approach where you write **specifications first**, then AI codes from them - Part of GitHub's spec-kit toolkit - Creates specs → plans → tasks → implementation **The Pattern:** - Write detailed spec docs upfront - AI generates implementation plan - Breaks into small, reviewable chunks - Each task implemented and tested in isolation **Use when:** You want to drive development from written specifications **Sources:** - [GitHub blog](https://github.blog/ai-and-ml/generative-ai/spec-driven-development-with-ai-get-started-with-a-new-open-source-toolkit/) - [Spec-driven guide](https://github.com/github/spec-kit/blob/main/spec-driven.md) --- ## Quick Comparison Table | Method | Focus | Best For | File Structure | |--------|-------|----------|----------------| | **planning-with-files** | Context engineering for Claude | Preventing AI goal drift during execution | 3 markdown files with hooks | | **TaskMaster AI** | PRD-based task management | Breaking down requirements into tasks | `.taskmaster/` folder with PRD + tasks | | **BMAD Method** | Agentic planning with specialized agents | Creating comprehensive specs upfront | YAML workflows + PRD + Architecture docs | | **Open Agent Spec** | Agent portability | Making agents work across frameworks | Agent definition configs | | **Spec-Driven Dev** | Specification-first coding | Driving implementation from detailed specs | Spec docs → plans → tasks | --- ## Why This Repo is Different After Manus sold for $2B, people realized their secret was **forcing AI to re-read goals constantly** to prevent context drift. This repo is the **only one** that: 1. Uses Claude Code v2.1+ hooks to **automate** the pattern 2. Makes the AI read `task_plan.md` **before every major decision** (PreToolUse hook) 3. **Blocks completion** until all phases are done (Stop hook) The others require **you** to manage the process. This one makes **Claude** manage it automatically. --- ## Try It ```bash /plugin marketplace add OthmanAdi/planning-with-files /plugin install planning-with-files@planning-with-files ``` Start a complex task and Claude will automatically handle the planning files without you telling it to. --- Hope this clears it up! The methods aren't really competing, they solve different problems. You could even use BMAD for upfront planning, then planning-with-files for execution tracking. And yeah, we should add Discussions! Good call. — Ahmad

No labels

bug

pull-request

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

starred/planning-with-files#10

No description provided.

Rows
Columns