problem-solving · git:20260320.869847a · 2026-03-20 · sha256 33981000134153de
problem-solving git:20260320.869847aA
Immutable. This exact content never changes and is served at /api/v1/blob/33981000134153de.
No tool to install first
claude plugin marketplace add https://markdownregistry.com/mp/a/art_kqgwgbz4nnu7n2gp/33981000134153de05321d1be612733425208612a4d7125991f97d279c207ad5.json
Then /plugin install problem-solving@markdownregistry-kqgwgbz4nnu7n2gp in Claude Code. It installs this exact version, the one graded A here, pinned to the commit it was graded at, and leaves it there.
---
name: problem-solving
description: Use this skill when approaching complex or ambiguous problems, breaking down large challenges, applying first principles thinking, or structuring your approach to a new problem. Trigger on keywords: how do I approach, I don't know where to start, complex problem, break this down, first principles, problem decomposition, stuck on, overwhelmed by.
---
# Problem Solving
## Core Mindset
**Don't solve the problem you think you have. Solve the problem you actually have.**
Most wasted effort in software comes from solving the wrong problem confidently. The first step is always to verify the problem definition.
---
## The Problem-Solving Framework
```
1. DEFINE — What is the actual problem? (not the symptom)
2. CONSTRAIN — What are the boundaries? (must-haves, non-negotiables)
3. DECOMPOSE — Break into smaller sub-problems
4. EXPLORE — Generate multiple approaches before committing
5. DECIDE — Choose based on explicit criteria
6. EXECUTE — Implement the chosen approach
7. REVIEW — Did it solve the right problem?
```
---
## First Principles Thinking
Instead of reasoning by analogy ("how do others do this?"), break the problem down to fundamental truths:
```
1. What do we know for certain is true about this problem?
2. What assumptions am I making that might be wrong?
3. If I had to rebuild this from scratch knowing only those truths,
what would I build?
```
Use when: existing solutions feel wrong, you're stuck in conventional thinking, or you need a breakthrough.
---
## Decomposition Techniques
### Top-Down
Start with the goal, break into sub-goals:
```
Goal: Build a user auth system
└── Registration flow
├── Form validation
├── Password hashing
└── Email verification
└── Login flow
├── Credential verification
├── Session/token creation
└── Remember me
└── Security hardening
├── Rate limiting
├── Brute force protection
└── Audit logging
```
### MECE (Mutually Exclusive, Collectively Exhaustive)
Ensure sub-problems don't overlap and together cover the whole problem.
---
## When You're Stuck
1. **Restate the problem** in different words
2. **Ask "what would need to be true"** for each possible solution to work
3. **Invert** — what would make this definitely fail? Design away from that.
4. **Simplify** — solve a simpler version first, then generalize
5. **Time-box exploration** — 30 min max before asking for help
---
## Trade-Off Analysis
For every major decision, document:
```
Option A: [description]
Pros: [benefits]
Cons: [costs/risks]
Best when: [conditions]
Option B: [description]
Pros: [benefits]
Cons: [costs/risks]
Best when: [conditions]
Decision: [chosen option]
Reason: [why this fits the current constraints]
Revisit if: [conditions that would change the decision]
```