Skip to content

A structured, Nobel Prize-based problem-solving workflow that helps you find the real problem before solving it. 🔍

Notifications You must be signed in to change notification settings

moazbuilds/reframe

Repository files navigation

A structured, Nobel Prize-based problem-solving workflow that helps you find the real problem before solving it. 🔍

Why

When you have a problem, you're trying to solve the story your brain already made up about it. You can't see your own blind spots. You might end up solving the story, not the situation.

Reframe is a single-agent workflow with 7 studied steps that consults you through finding and solving the real problem.

The 7 Steps

  • Step 1 · Surface Get the real facts and feelings out. Not the story, the raw material.
  • Step 2 · Untangle Separate the threads. Most problems are multiple problems wearing one name.
  • Step 3 · Diagnose Find the root cause, not the symptom that's screaming loudest.
  • Step 4 · Challenge Stress-test the diagnosis. If it breaks, better now than later.
  • Step 5 · Design Build a concrete plan with specific actions, not intentions.
  • Step 6 · Act One first step. Not a to-do list. One thing you do now.
  • Step 7 · Check Did it work? What did you learn? Where else does this apply?

Installation

CodeMachine CLI (Recommended)

Make sure you have CodeMachine CLI installed.

codemachine import moazbuilds/reframe

Then choose it from /templates and run it with /start.

The default engine is Claude. Make sure it's authenticated using /login before you run /start.

NPX Skills

npx skills add https://github.com/moazbuilds/reframe

Then use it inside Claude Code, or any CLI tool or IDE that accepts skills, by running /reframe and following the workflow.

How It Works

You talk to Lens ꒰ᐡ •̥ ̫ •̥ ᐡ꒱, a problem-solving consultant persona that guides you through the 7 steps. She asks one question at a time, mirrors back what she hears, and separates facts from opinions.

  • The agent uses chained prompts. Each step loads its own specialized prompt, one after another, so the agent consults you like an expert would
  • Each step has built-in depth tracking. The agent knows when to dig deeper and when a step is complete
  • Every prompt follows strict prompt engineering principles. The model doesn't guess. It follows the framework and adapts to your specific case

Built On

  • George Polya (Structured Problem-Solving) The 4-phase structure: Understand → Plan → Execute → Review
  • Edward de Bono (Lateral Thinking) Challenge step stress-tests assumptions from multiple angles
  • Daniel Kahneman (Behavioral Economics) Addresses cognitive biases: narrative fallacy, overconfidence, substitution

About

A structured, Nobel Prize-based problem-solving workflow that helps you find the real problem before solving it. 🔍

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors