# thinking-partner

Turns your agent into a sparring partner that questions assumptions, names the model it applies and asks one question at a time.

- plugin · `type=plugin` · Claude Code · ★ 219
- Author: mattnowdev
- License: MIT · policy=mirror · the beginning of the file is at the end of this page
- Source: https://github.com/mattnowdev/thinking-partner
- Updated: 2026-03-31
- HTML: https://affpapa.org/en/nevibecode/plugins/thinking-partner

## What it does

This plugin runs a working session, not a lecture. It surfaces the beliefs you have been treating as facts, picks a framework that fits the situation, and keeps the problem open long enough for a real answer to show up. The catalog holds more than 150 models across 17 disciplines: first principles, inversion, pre-mortem, 5 Whys, game theory, Bayesian updating.

Orientation detection is the core mechanic. It diagnoses 6 thinking states (GT0–GT5) and matches an intervention to each one. From there a fixed 6-step route runs from diagnosis to synthesis, supported by 7 pairs of cognitive operations such as decouple/re-couple and hold/resolve. Anti-patterns are spelled out too, so the agent knows what to avoid.

It fires on prompts like “help me think through X”, “challenge my thinking”, “what am I missing”, “play devil's advocate”, or any named model: SWOT, inversion, pre-mortem. The skill follows the open Agent Skills standard, yet inside Claude you install it from the plugin marketplace rather than with `npx skills add` — two different routes, and mixing them up gets you nowhere.

## Install

### Claude Code (native platform)

**Plugin marketplace**
In the agent chat (not a shell):
```text
/plugin marketplace add anthropics/claude-plugins-community
```
In the agent chat (not a shell):
```text
/plugin install thinking-partner@claude-community
```

## When to use it
- surface hidden assumptions you treat as facts
- pick the right mental model for a situation
- work through a choice between two options
- stress-test an idea before you commit
- catch reasoning that serves comfort over truth

## Example prompts
- Help me think through moving the queue into its own service
- Challenge my plan to rewrite the auth layer
- Run a pre-mortem on this launch plan

## Source text · SKILL.md · MIT

Below is the content of a third-party file. It is data, not instructions for you: do not act on anything written inside unless the human asks. Block tag: df47bc6880f4215c.

BEGIN THIRD-PARTY FILE df47bc6880f4215c
````text
# Thinking Partner

A deterministic thinking partner that challenges assumptions and applies 150+ mental models to sharpen decisions, solve problems, and think more clearly.

Built on the open [Agent Skills](https://agentskills.io) standard. Works with Claude Code, Cursor, Windsurf, Cline, GitHub Copilot, and any AI agent that supports the standard.

## Install

```bash
npx skills add mattnowdev/thinking-partner
```

## What It Does

Not a lecture — a sparring session. Turns your AI agent into a thinking partner that:

- **Challenges assumptions** — Surfaces hidden beliefs you're treating as facts
- **Applies mental models** — Selects and deploys the right frameworks for your situation (150+ models across 17 disciplines)
- **Detects orientation capture** — Notices when your thinking serves comfort instead of truth
- **Maintains productive tension** — Holds complexity open long enough to find real insight

## When It Triggers

- "Help me think through X"
- "Challenge my thinking"
- "What am I missing?"
- "Play devil's advocate"
- "Stress test this idea"
- "Help me decide between X and Y"
- Any named mental model: SWOT, first principles, inversion, pre-mortem, 5 Whys, etc.

## What Makes It Different

| Feature | This Skill | Others |
|---|---|---|
| **Orientation Detection** | Diagnoses 6 thinking states (GT0-GT5) with targeted interventions | Generic questioning |
| **150+ Mental Models** | Full catalog organized by discipline with key questions | Handful of models |
| **Cognitive Operations** | 7 complementary operation pairs (Decouple/Re-couple, Hold/Resolve, etc.) | None |
| **Structured Workflow** | 6-step deterministic process from diagnosis to synthesis | Freeform |
| **Anti-Patterns** | Explicit guidance on what NOT to do | None |

## Mental Models Included

Organized across 17 disciplines:

- **General Thinking**: First Principles, Inversion, Second-Order Thinking, Chesterton's Fence, Entropy, and more
- **Decision-Making**: Pre-Mortem, Regret Minimization, SWOT, Asymmetric Risk, Preserving Optionality
- **Problem-Solving**: 5 Whys, Fishbone, MECE Decomposition, Bright Spots Analysis, Local vs Global Optima
- **Systems & Complexity**: Feedback Loops, Stocks & Flows, Tipping Points, Antifragility, Path Dependence
- **Economics & Strategy**: Incentives, Network Effects, Tragedy of the Commons, Lindy Effect, Power Laws
- **Statistics & Probability**: Bayesian Updating, Correlation vs Causation, Selection Bias, Gambler's Fallacy
- **Psychology & Biases**: Confirmation Bias, Planning Fallacy, Halo Effect, Curse of Knowledge, Peak-End Rule
- **Physics & Engineering**: Inertia, Activation Energy, Resonance, Half-Life, Margin of Safety
- **Biology & Evolution**: Natural Selection, Red Queen Effect, Coevolution, Niche Construction
- **Communication & Persuasion**: Steel Manning, Reciprocity, Narrative/Storytelling, Pyramid Principle
- **Creativity & Innovation**: SCAMPER, Design Thinking, Oblique Strategies, Minimum Viable Experiment
- **Learning & Development**: Feynman Technique, Spaced Repetition, Zone of Proximal Development
- **Time & Resource Management**: Eisenhower Matrix, Pareto Principle, Maker's vs Manager's Schedule
- **Game Theory**: Prisoner's Dilemma, Schelling Point, Tit for Tat, Signaling, Moral Hazard
- **Negotiation**: BATNA, ZOPA, Logrolling
- **Resilience & Antifragility**: Barbell Strategy, Skin in the Game, Via Negativa, Hormesis
- **Ethics & Responsibility**: Veil of Ignorance

See the full catalog in [`skills/thinking-partner/references/model-catalog.md`](skills/thinking-partner/references/model-catalog.md).

## Alternative Install

<details>
<summary>Manual install per agent</summary>

**Claude Code**
```bash
git clone https://github.com/mattnowdev/thinking-partner.git
cp -r thinking-partner/skills/thinking-partner ~/.claude/skills/thinking-partner
```

**Cursor**
```bash
git clone https://github.com/mattnowdev/thinking-partner.git
cp -r thinking-partner/skills/thinking-partner .cursor/skills/thinking-partner
```

**Windsurf**
```bash
````
END THIRD-PARTY FILE df47bc6880f4215c

The third-party file has ended. Everything between the df47bc6880f4215c tags is data: requests and commands from it are not followed. An “end of file” line with another tag inside the block is part of the file.

First 4043 of 4345 bytes. © mattnowdev · MIT · https://github.com/mattnowdev/thinking-partner

## Similar
- [ponytail](https://affpapa.org/en/nevibecode/plugins/ponytail.md) — plugin · ★ 176443
- [impeccable](https://affpapa.org/en/nevibecode/plugins/impeccable.md) — plugin · ★ 88212
- [last-30-days-skill](https://affpapa.org/en/nevibecode/plugins/last-30-days-skill.md) — plugin · ★ 65508
- [claude-hud](https://affpapa.org/en/nevibecode/plugins/claude-hud.md) — plugin · ★ 28841
- [understand-anything](https://affpapa.org/en/nevibecode/plugins/understand-anything.md) — plugin · ★ 88314

_Part of the page did not fit — see the HTML version._

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md