ManagerMirror

🇺🇸 English | 🇰🇷 한국어

A structured coaching framework for anyone navigating the manager-engineer relationship — whether you're the one managing or the one being managed.

The hardest part of working in tech isn't the code. It's the people. This project helps you work through real situations, understand the other side's perspective, and build a personal philosophy — one reflection at a time.

Originally built for engineering managers. Now equally useful for engineers trying to understand their managers, navigate difficult dynamics, or process experiences like PIP, performance reviews, or unexpected conflicts.

---

Why This Exists

The manager-engineer relationship is one of the most consequential — and least understood — dynamics in tech. From the manager's side: giving feedback, handling underperformance, building trust. From the engineer's side: navigating PIPs, understanding why a manager reacted the way they did, or learning how organizational power actually works.

Both sides often lack a structured space to process what happened and build genuine understanding of the other.

ManagerMirror provides:

• A knowledge base of foundational frameworks (psychology, feedback, trust, leadership)
• A structured format for recording and reflecting on real situations
• A coaching approach based on the GROW model — surfaces questions rather than prescribing answers
• Perspective-taking — when you can't understand the other side, the skill generates plausible hypotheses and explores them with you
• Multi-AI support — works with Claude Code, GitHub Copilot, ChatGPT, Hermes (Ollama), or any AI that can read files

---

How It Works

Sessions follow the GROW model, extended with a Perspective-Taking stage:

Real situation
      ↓
G — Goal: what do you want from this session?
      ↓
R — Reality: what is actually happening? (one question at a time)
      ↓
P — Perspective: what might the other side be thinking, feeling, fearing, needing?
      ↓
O — Options: what could you do differently?
      ↓
W — Will: what will you actually do?
      ↓
Principles — accumulated over time

Each session produces three files:

• situation.md — what happened, in facts only
• challenge.md — the uncomfortable questions raised
• insight.md — what was realized, what changes next time

Patterns accumulate in profile/. Principles accumulate in principles/. Over time, the mirror gets sharper.

---

Getting Started

git clone https://github.com/{your-username}/ManagerMirror.git
cd ManagerMirror

Then start a session with your AI tool of choice. See Supported AI Tools below.

Your personal folder (users/{your_username}/) is created automatically as sessions progress — no manual setup needed.

Your users/ folder is listed in .gitignore and will never be committed. Your data stays local.

---

Example Session

Type /managermirror, then describe your situation. Here's how a session looks from each side:

	👨‍💻 Engineer	👔 Manager
You describe the situation	"I got 'meets expectations' in my annual review. My manager never raised any concerns in our 1:1s all year. I don't know what I missed."	"I gave my engineer critical feedback in our 1:1 about their communication. They've barely spoken to me since. I don't know if I handled it wrong."
Claude — Reality	"Looking back at your 1:1s this past year: were there any moments where your manager seemed hesitant, gave softer feedback than you expected, or changed the subject?"	"What exactly did you say? As close to your actual words as you can."
Claude — Perspective	"What might have constrained your manager from raising concerns directly — organizationally or personally? (See foundations/12)"	"Your engineer went quiet after the feedback. What might they be feeling right now — and what might they be afraid of?"
Session output	situation.md · challenge.md · insight.md	situation.md · challenge.md · insight.md

Each stage surfaces one question at a time. The session ends when you reach a concrete next action — or a principle worth keeping.

---

Project Structure

ManagerMirror/
├── .gitignore              # excludes users/
├── AGENTS.md               # instructions for all AI tools
├── README.md
├── README.ko.md
├── skill/
│   └── SKILL.md            # Claude Code skill definition
├── foundations/            # English knowledge base — primary (AI reads this)
│   ├── 01_human_is_not_a_system.md
│   ├── 02_basic_human_needs.md
│   ├── ...
│   └── 14_manager_managing_engineers.md
├── foundations.ko/         # Korean knowledge base — user reference only (Korean filenames)
│   ├── 01_사람은_시스템이_아니다.md
│   ├── 02_기본_심리_욕구.md
│   ├── ...
│   └── 14_매니저의_엔지니어_관리.md
├── template/               # file format references
│   ├── situation.md
│   ├── challenge.md
│   ├── insight.md
│   ├── patterns.md
│   └── principles.md
└── users/                  # gitignored — your data only
    └── {your_username}/
        ├── profile/
        ├── situations/
        └── principles/

---

Foundations Knowledge Base

The foundations/ folder contains 14 reference documents that power the coaching sessions:

File	Topic
01	Why people are not systems — the engineer's mental model trap
02	Self-determination theory — autonomy, competence, relatedness
03	Psychological safety — the single biggest predictor of team performance
04	The trust equation — why self-orientation destroys trust
05	Radical Candor — care personally, challenge directly
06	Nonviolent Communication — observation → feeling → need → request
07	Difficult conversations — the 3 layers every hard talk contains
08	Feedback — SBI model, timing, and how to receive it
09	Situational leadership — matching your style to the person's stage
10	Engineering manager traps — fix-it mode, clarity obsession, and more
11	Questioning techniques — how to draw out what people can't say yet
12	What managers can't say — legal, HR, organizational, and psychological constraints that explain silence and indirect communication
13	Engineer managing up — visibility, 1:1s, feedback channels, career advocacy, escalation
14	Manager managing engineers — engineer mindset, deep work, seniority calibration, early feedback

---

Supported AI Tools

ManagerMirror works with any AI that can read files. Full instructions are in AGENTS.md.

AI Tool	How to use
Claude Code	Install skill/SKILL.md → type /managermirror
GitHub Copilot	Open project in VS Code → @workspace with AGENTS.md
ChatGPT	Upload AGENTS.md + relevant foundations files
Hermes (Ollama)	Use AGENTS.md as system prompt

Claude Code (full automation)

Copy the skill to your Claude skills directory:

# macOS
cp -r skill/ ~/Library/Application\ Support/Claude/claude_desktop_config/skills/managermirror/

# Windows
cp -r skill/ "$env:APPDATA\Claude\skills\managermirror\"

Then in any Claude Code session:

/managermirror

The skill auto-discovers the project location and your username — no configuration needed.

Other AI tools

See AGENTS.md for step-by-step instructions for each tool.

---

Privacy Model

What	Where	Committed?
Knowledge base	foundations/	✅ Yes
Templates	template/	✅ Yes
Your situations	users/{you}/situations/	❌ No
Your profile	users/{you}/profile/	❌ No
Your principles	users/{you}/principles/	❌ No

You can share the framework without sharing any of your personal data.

---

License

MIT