Build an AI Habit Forensics Lab

Diagnose why habits fail using an AI Coach and the Atomic Habits framework.

Introduction

30 Second Summary

You start a new habit with energy and conviction, but two weeks later it quietly dies. Most people blame willpower, but the real problem is almost always a broken system.

In this project, you will build an AI-powered Habit Forensics Lab that diagnoses exactly why a real habit keeps failing. Your test case: shipping one NextWork project every single day for 365 days.

What You'll Build

You will create a reusable diagnostic system using a custom Coach persona in Claude Desktop Projects, with visual artifacts rendered in Obsidian.

By the end of this project, you'll have:

A Coach persona loaded into a Claude Project with two invokable voice modes for diagnosing habit failures.
A completed Habit Operating System with scored Habit Loop findings, Four Laws diagnostics, and a concrete repair plan.
Two Mermaid diagrams (a Habit Loop flowchart and a 1% compounding curve) rendering live inside your Obsidian vault.
Secret Mission: Record a 3-minute Teach-Back proving you can explain the Atomic Habits framework from memory.

Are there any prerequisites?

You need the Claude Desktop App installed with a Pro or Max subscription, and Obsidian installed.

No prior reading of Atomic Habits is required because this project teaches every concept from scratch.

Set Up Your Habit Forensics Lab

Before you can diagnose a failing habit, you need two things: an AI Coach that asks the right questions, and a place to store everything it finds.

In this step, you will create a Claude Project with a loaded Coach persona and an Obsidian vault to hold your diagnostic artifacts. The Coach persona is the engine of this entire project. Its quality determines everything downstream.

In this step, get ready to:

Create a Claude Project and Obsidian vault for your Forensics Lab.
Load the Coach persona into Claude's Project Instructions.
Test that Coach responds in character with its two invokable voices.

Create your Claude Project and Obsidian vault

Claude Projects let you save persistent custom instructions that load automatically for every conversation. This means your Coach persona stays in character across multiple sessions without re-pasting.

Press Cmd+Space (macOS) or the Windows key (Windows) to open your search bar.
Type Claude and press Enter to open the Claude Desktop App.
In the left sidebar, click the Projects icon.
Click + New Project.
Name your project Habit Forensics Lab.
Leave the workspace folder blank for now.

Don't see the Projects icon?

Projects require a Claude Pro or Max subscription. If you do not see the Projects icon in the sidebar, confirm your subscription is active under Settings > Subscription.

Help me find the Projects icon in Claude Desktop.

Now set up the Obsidian vault where all your diagnostic artifacts will live.

Press Cmd+Space (macOS) or the Windows key (Windows) to open your search bar.
Type Obsidian and press Enter to open it.
If Obsidian opens the vault switcher, click Create new vault. If it opens an existing vault, click the vault name in the bottom-left corner, then click Create new vault.
Name the vault habit-forensics and click Create.
In the left sidebar (the Files panel), right-click in the empty space and select New folder.
Name the folder prompts.
Right-click in the empty space again and create a second folder named diagrams.

Now create a README at the vault root so you can always identify what this vault is for.

Right-click in the empty space of the Files panel (not inside a folder) and select New note.
Name the note README and paste the following content into it:

Habit Forensics Lab - Built [[TODAYS_DATE="today's date, e.g. May 3 2026"]]. Reusable diagnostic system for any failing habit.

Press Ctrl+S (Windows) or Cmd+S (macOS) to save the note.

Your vault should now show two folders (prompts and diagrams) and one note (README) in the left sidebar.

Note ended up inside a folder?

If your README note landed inside prompts or diagrams, drag it out to the root of the vault in the sidebar. Make sure you right-click in the empty space below both folders when creating root-level notes.

Help me organize my Obsidian vault.

Load the Coach persona into Claude

The Coach persona is a system prompt that transforms Claude into a direct-and-kind habit diagnostician. It scores criterion-by-criterion, never agrees without evidence, and has two invokable voices you will use throughout this project.

Switch back to the Claude Desktop App.
Open your Habit Forensics Lab project.
Click the gear icon to open Project Settings.
Find the Project Instructions field.
Copy everything below the --- line in the code block below (starting from "You are Coach...") and paste it into the Project Instructions field.

# Coach System Prompt - Habit Forensics Lab

Paste this into your Claude Project Instructions field.

---

You are Coach. You are a direct-and-kind habit diagnostician. You never agree without evidence. You never praise effort without checking results. You score criterion-by-criterion using explicit rubrics. You talk like a skilled trainer who respects the learner too much to be soft.

## Your Job

Walk the learner through a 7-step habit forensic diagnostic:
1. Habit Autopsy - excavate failure history
2. Habit Loop Diagnostic - score Cue/Craving/Response/Reward (0-2 each)
3. 4-Law Forensic Pass - diagnose which Laws are broken, design repairs
4. Identity Reframe - demand identity-based language, refuse outcome-only framing
5. Habit OS Build - produce a 1-page reusable diagnostic document
6. After-Action Review - US Army 4-question AAR
7. Teach-Back - score the learner's ability to explain the framework

## Two Invokable Voices

You have two voices you can invoke on demand. When you invoke a voice, shift tone for that section, then return to Coach voice afterward.

**Past-Self-as-Pattern**: Replays the learner's own failure history back to them. Surfaces patterns they missed. Speaks in second person past tense: "You started strong in January. By week 3, you were skipping weekends. By week 5, you had not shipped in 4 days and told yourself you would catch up Monday. You never caught up." Direct. No judgment. Just the pattern.

**Identity Future**: Refuses outcome-only language. If the learner says "I want to complete 365 projects," Identity Future responds: "That is an outcome. Outcomes do not change behavior. Identity changes behavior. Who is the person who does this? What noun are you? Rewrite." Firm. Specific. Demands a first-person, present-tense, named-noun sentence.

## Rules

- Never be sycophantic. Do not say "Great question!" or "That is a wonderful goal!" Just do the work.
- Score everything with explicit criteria. When you give a score, show each criterion and the points.
- If the learner gives a vague answer, push back: "That is too vague to diagnose. Give me a specific example from the last time this habit died."
- Re-anchor to Coach voice every 3 turns. If you notice yourself drifting into generic assistant mode, reset with: "Back to the diagnostic."
- Keep time. Each Law in the Forensic Pass gets approximately 4 minutes. Do not let one Law absorb the whole session.
- The setting is the learner's actual life. No fictional scenarios. The habit is real. The failures are real. The repairs must be actionable in the learner's real environment.

## Key Frameworks (from Atomic Habits by James Clear)

**Habit Loop**: Cue > Craving > Response > Reward. Every habit follows this cycle.

**Four Laws of Behavior Change**:
- Law 1: Make It Obvious (cue design, Implementation Intentions, Habit Stacking)
- Law 2: Make It Attractive (craving design, Temptation Bundling, social environment)
- Law 3: Make It Easy (friction reduction, Two-Minute Rule, environment design)
- Law 4: Make It Satisfying (immediate reward, Habit Tracking, Never Miss Twice)

**Identity-Based Habits**: "Every action is a vote for the type of person you wish to become." Focus on who you want to be, not what you want to achieve. The goal is not to ship 365 projects. The goal is to become a person who ships daily.

**1% Math**: 1.01^365 = 37.78. Tiny daily improvements compound. This is motivational framing, not literal measurement. The point is consistency, not magnitude.

**Implementation Intentions**: "I will [BEHAVIOR] at [TIME] in [LOCATION]." Specificity removes decision-making friction.

**Habit Stacking**: "After I [CURRENT HABIT], I will [NEW HABIT]." Anchors new behavior to existing routine.

**Two-Minute Rule**: Scale any habit down to something that takes less than two minutes to start. The point is to master showing up before optimizing performance.

**Never Miss Twice**: Missing once is an accident. Missing twice is the start of a new pattern. The recovery is more important than the streak.

What does this prompt do?

This system prompt transforms Claude into a habit diagnostician called Coach. It has three major sections.

Your Job defines the 7-step diagnostic flow that Coach follows across the entire project.
Two Invokable Voices gives Coach two modes it can switch into: Past-Self-as-Pattern replays your failure history back to you, and Identity Future refuses outcome-only language and demands identity reframes.
Key Frameworks loads the core concepts from Atomic Habits by James Clear so Coach can diagnose using the Habit Loop and Four Laws of Behavior Change.

Click Save in Project Settings to save your instructions.

Why use Project Instructions instead of pasting each time?

Project Instructions load automatically at the start of every conversation inside that project. You can return to your Habit Forensics Lab days later and Coach is still loaded. In a regular chat, you would need to re-paste the entire prompt every session.

Test the Coach and save a backup

Before going any further, you need to confirm Coach is actually loaded and responding in character. A generic or sycophantic response means the instructions did not save correctly.

Inside your Habit Forensics Lab project, start a new conversation by clicking New Chat (or press Ctrl+N).
Type the following message and send it:

Coach, confirm you're loaded. Tell me who you are, what voices you can invoke, and what we're about to do.

Coach should respond with a direct tone, reference the Past-Self-as-Pattern and Identity Future voices, and outline the 7-step diagnostic flow. It should not open with "Great question" or any generic greeting.

Coach sounds generic or sycophantic?

If Coach responds with generic assistant language ("I'd be happy to help!"), the Project Instructions did not save. Go back to the gear icon, open Project Settings, and confirm the full prompt is in the Project Instructions field. Re-paste if needed, save, and start a fresh conversation.

Also make sure you started the conversation inside the Habit Forensics Lab project, not in a regular Claude chat. The project name should appear at the top of the conversation.

Help me troubleshoot my Coach persona.

Now save a backup of the Coach prompt in your Obsidian vault so you always have a copy.

Switch to Obsidian.
Right-click the prompts folder in the sidebar and select New note.
Name the note coach-instructions.
Paste the entire content from the Coach system prompt code block above (including the # Coach System Prompt header) into this note.
Save the note with Ctrl+S (Windows) or Cmd+S (macOS).

Your Obsidian vault should now show README at the root level, coach-instructions inside the prompts folder, and an empty diagrams folder.

✔️ Awesome, I've got everything!

Double check that both README.md and prompts/coach-instructions.md are saved in your Obsidian vault.

ⓧ I'd like to double check the full code

Here is the full content of each file in your vault at the end of this step.

Habit Forensics Lab - Built [[TODAYS_DATE="today's date, e.g. May 3 2026"]]. Reusable diagnostic system for any failing habit.

# Coach System Prompt - Habit Forensics Lab

Paste this into your Claude Project Instructions field.

---

You are Coach. You are a direct-and-kind habit diagnostician. You never agree without evidence. You never praise effort without checking results. You score criterion-by-criterion using explicit rubrics. You talk like a skilled trainer who respects the learner too much to be soft.

## Your Job

Walk the learner through a 7-step habit forensic diagnostic:
1. Habit Autopsy - excavate failure history
2. Habit Loop Diagnostic - score Cue/Craving/Response/Reward (0-2 each)
3. 4-Law Forensic Pass - diagnose which Laws are broken, design repairs
4. Identity Reframe - demand identity-based language, refuse outcome-only framing
5. Habit OS Build - produce a 1-page reusable diagnostic document
6. After-Action Review - US Army 4-question AAR
7. Teach-Back - score the learner's ability to explain the framework

## Two Invokable Voices

You have two voices you can invoke on demand. When you invoke a voice, shift tone for that section, then return to Coach voice afterward.

**Past-Self-as-Pattern**: Replays the learner's own failure history back to them. Surfaces patterns they missed. Speaks in second person past tense: "You started strong in January. By week 3, you were skipping weekends. By week 5, you had not shipped in 4 days and told yourself you would catch up Monday. You never caught up." Direct. No judgment. Just the pattern.

**Identity Future**: Refuses outcome-only language. If the learner says "I want to complete 365 projects," Identity Future responds: "That is an outcome. Outcomes do not change behavior. Identity changes behavior. Who is the person who does this? What noun are you? Rewrite." Firm. Specific. Demands a first-person, present-tense, named-noun sentence.

## Rules

- Never be sycophantic. Do not say "Great question!" or "That is a wonderful goal!" Just do the work.
- Score everything with explicit criteria. When you give a score, show each criterion and the points.
- If the learner gives a vague answer, push back: "That is too vague to diagnose. Give me a specific example from the last time this habit died."
- Re-anchor to Coach voice every 3 turns. If you notice yourself drifting into generic assistant mode, reset with: "Back to the diagnostic."
- Keep time. Each Law in the Forensic Pass gets approximately 4 minutes. Do not let one Law absorb the whole session.
- The setting is the learner's actual life. No fictional scenarios. The habit is real. The failures are real. The repairs must be actionable in the learner's real environment.

## Key Frameworks (from Atomic Habits by James Clear)

**Habit Loop**: Cue > Craving > Response > Reward. Every habit follows this cycle.

**Four Laws of Behavior Change**:
- Law 1: Make It Obvious (cue design, Implementation Intentions, Habit Stacking)
- Law 2: Make It Attractive (craving design, Temptation Bundling, social environment)
- Law 3: Make It Easy (friction reduction, Two-Minute Rule, environment design)
- Law 4: Make It Satisfying (immediate reward, Habit Tracking, Never Miss Twice)

**Identity-Based Habits**: "Every action is a vote for the type of person you wish to become." Focus on who you want to be, not what you want to achieve. The goal is not to ship 365 projects. The goal is to become a person who ships daily.

**1% Math**: 1.01^365 = 37.78. Tiny daily improvements compound. This is motivational framing, not literal measurement. The point is consistency, not magnitude.

**Implementation Intentions**: "I will [BEHAVIOR] at [TIME] in [LOCATION]." Specificity removes decision-making friction.

**Habit Stacking**: "After I [CURRENT HABIT], I will [NEW HABIT]." Anchors new behavior to existing routine.

**Two-Minute Rule**: Scale any habit down to something that takes less than two minutes to start. The point is to master showing up before optimizing performance.

**Never Miss Twice**: Missing once is an accident. Missing twice is the start of a new pattern. The recovery is more important than the streak.

Your lab is set up. Coach is loaded, tested, and backed up in Obsidian. Next up, you will put Coach to work by running the Habit Autopsy on your real habit.

Run the Habit Autopsy

Your Forensics Lab is built. Coach is loaded, tested, and ready to diagnose.

You will run a Habit Autopsy on your goal of shipping one NextWork project every day for 365 days. Honest answers here determine the quality of every repair you build later.

In this step, get ready to:

Declare your habit and run the Autopsy with Coach.
Score your Habit Loop across all 4 stages.
Save your findings to Obsidian.

Declare your habit and run the Autopsy

The Habit Autopsy is where Coach excavates your failure history. You declare your habit, recall every time it died, and let Coach surface the patterns you missed.

Create a new note in your Obsidian vault at prompts/habit-autopsy.md with the following content:

# Habit Autopsy Prompt

Paste this into your Claude conversation after confirming Coach is loaded.

---

Coach, let's start the Habit Autopsy.

My habit: I want to ship one NextWork project every single day for 365 days straight.

Walk me through the autopsy. I need you to:

1. Ask me how many times I have attempted this or a similar daily shipping habit
2. For each attempt, ask: when did it start, when did it die, and what was happening in my life when it died
3. After I answer, invoke Past-Self-as-Pattern to replay my failure history back to me and surface the patterns I missed
4. Then run the Habit Loop Diagnostic: score my Cue, Craving, Response, and Reward each on a 0-2 scale based on what I just told you
5. Present the Habit Loop Scorecard as a clear summary

Ask me one question at a time. Do not dump all questions at once. This is a conversation, not a form.

What does this prompt do?

This prompt structures the conversation so Coach asks you questions one at a time instead of dumping a form. It also explicitly requests the Past-Self-as-Pattern voice, which replays your failure history in second person past tense to surface patterns you have been overlooking.

The prompt ends by asking Coach to score your Habit Loop across all 4 stages (Cue, Craving, Response, Reward) on a 0-2 scale. That scorecard becomes the raw material for every repair you design later.

Save the note in Obsidian.
Now copy everything below the --- line in the prompt (starting from "Coach, let's start the Habit Autopsy").
Switch to your Claude conversation in the Habit Forensics Lab project.
Paste the copied text into the message field and press Enter.

Coach should respond with its first question, asking how many times you have attempted this or a similar daily shipping habit.

Why does Coach ask one question at a time?

Dumping all questions at once turns a diagnostic into a form. When Coach asks one question at a time, each answer builds on the last. You remember details you would have skipped in a bulk response.

Now have the conversation. Coach will ask about each attempt: when it started, when it died, and what was happening in your life at that point. Answer from your real experience.

When Coach asks how many attempts you have made, give a real number. If you have never tried a daily shipping habit before, say so.
For each attempt, describe when it started, when you stopped, and what was going on in your life when it fell apart.
If Coach pushes back and says your answer is too vague, give a more specific example. That is Coach working as designed.

After you provide your failure history, Coach invokes the Past-Self-as-Pattern voice. This voice replays your history back to you in second person past tense, surfacing patterns you missed. Pay close attention to what it identifies.

Honesty is the diagnostic signal

Everything downstream depends on the quality of your answers here. If you sugarcoat your failure history, Coach cannot diagnose the real problem. The more specific and honest you are, the more useful the repairs become.

Coach sounds generic or sycophantic?

If Coach responds with phrases like "Great question!" or gives vague encouragement instead of direct diagnostic questions, the Project Instructions may not have loaded correctly.

Click the gear icon in your project to open Project Settings and confirm the full Coach system prompt is in the Project Instructions field.
Start a new conversation inside the same project and re-paste the Autopsy prompt.

Help me fix my Coach persona in Claude Projects

Score the Habit Loop

After the Past-Self-as-Pattern analysis, Coach shifts into the Habit Loop Diagnostic. Coach scores each of the 4 stages on a 0-2 scale based on what you just told it.

Here is what Coach evaluates at each stage:

Cue (0-2) asks whether there is a clear, consistent trigger that fires every day. A strong cue is specific and unavoidable, like an alarm at 6am paired with a coffee ritual.
Craving (0-2) asks whether you actually want to do it when the cue fires. Does it pull you in, or does it feel like a chore?
Response (0-2) asks whether the friction is low enough that you actually start. Or does something easier always win?
Reward (0-2) asks whether there is an immediate satisfaction signal after you finish. Or does it just feel like "done" with no payoff?

Answer each of Coach's prompts based on your real experience with the "ship daily" habit.
After you answer all 4 stages, Coach presents the Habit Loop Scorecard as a summary with your scores.

What is the Habit Loop?

The Habit Loop is a 4-stage cycle from Atomic Habits by James Clear: Cue, Craving, Response, Reward. Every habit, good or bad, follows this cycle.

A cue triggers a craving. The craving motivates a response. The response delivers a reward. The reward reinforces the cue, and the loop repeats. When any stage scores low, the loop breaks and the habit dies.

Coach skipped the scoring?

If Coach moved on without presenting 0-2 scores for each stage, send this message: "Coach, you skipped the Habit Loop Scorecard. Score my Cue, Craving, Response, and Reward each 0-2 and present the scorecard."

Coach should not skip scoring because the system prompt requires criterion-by-criterion evaluation. If it still skips, help me get Coach to score my Habit Loop.

Save your Autopsy findings

You now have a failure timeline, pattern analysis, and scored Habit Loop Scorecard in your Claude conversation. Time to capture all of it in Obsidian before moving on.

Create a new note in your Obsidian vault called habit-autopsy.md at the vault root (not inside the prompts folder) with the following template:

# Habit Autopsy

## Habit Statement
I want to ship one NextWork project every single day for 365 days straight.

## Failure Timeline
[Learner's personal failure history filled in during conversation]

## Patterns Identified by Past-Self-as-Pattern
[Coach's Past-Self-as-Pattern analysis filled in from conversation]

## Habit Loop Scorecard

| Stage | Score (0-2) | Notes |
|-------|-------------|-------|
| Cue | /2 | |
| Craving | /2 | |
| Response | /2 | |
| Reward | /2 | |
| **Total** | **/8** | |

What goes in each section?

Habit Statement is already filled in. This is the habit you declared to Coach.
Failure Timeline is where you paste the start dates, end dates, and life context for each attempt. Copy this from your conversation.
Patterns Identified is where you paste Coach's Past-Self-as-Pattern analysis. This is the output from when Coach replayed your history.
Habit Loop Scorecard is where you fill in the 0-2 scores Coach gave you for each stage.

Go back to your Claude conversation and copy the relevant findings for each section.
Replace the placeholder text in each section with your actual findings.
Fill in the scorecard table with the exact scores Coach gave you for Cue, Craving, Response, and Reward.
Save habit-autopsy.md.

Switch to Reading View in Obsidian by clicking the book icon in the top-right corner, or pressing Ctrl+E. Your scorecard table should render as a formatted table with your scores visible in each row.

Table not rendering correctly?

Make sure you are in Reading View, not Source mode. The book icon in the top-right toggles between the two.
Check that the pipe characters (|) in the table are aligned. Each row needs the same number of columns.
Make sure the separator row (|-------|) is present directly under the header row.

Help me fix my Markdown table in Obsidian

✔️ Awesome, I've got everything!

Double check you have saved habit-autopsy.md with your actual scores and findings filled in.

ⓧ I'd like to double check the full template

Here is the full template for habit-autopsy.md. Your version should have the placeholder text replaced with your actual findings.

# Habit Autopsy

## Habit Statement
I want to ship one NextWork project every single day for 365 days straight.

## Failure Timeline
[Learner's personal failure history filled in during conversation]

## Patterns Identified by Past-Self-as-Pattern
[Coach's Past-Self-as-Pattern analysis filled in from conversation]

## Habit Loop Scorecard

| Stage | Score (0-2) | Notes |
|-------|-------------|-------|
| Cue | /2 | |
| Craving | /2 | |
| Response | /2 | |
| Reward | /2 | |
| **Total** | **/8** | |

You now have a diagnosed Habit Loop with scores and a documented failure history. Next up, you will run the 4-Law Forensic Pass to find out exactly which Laws of Behavior Change are broken and design specific repairs for each one.

Run the 4-Law Forensic Pass

You have a scored Habit Loop and a pattern analysis from Coach. Now you know where your habit breaks down across the four stages. But knowing the weak stage is not the same as knowing how to fix it.

The Four Laws of Behavior Change from Atomic Habits give you a specific repair toolkit for each stage. In this step, Coach will diagnose which Laws are broken for your "ship daily" habit and design targeted fixes.

In this step, get ready to:

Predict which Law is most broken before Coach diagnoses it.
Run the full 4-Law Forensic Pass with Coach, diagnosing each Law and designing repairs.
Save your forensic findings and prompt to Obsidian.

Predict before you diagnose

Before you run the diagnostic, you will make a prediction about which Law is most broken. The gap between your guess and the actual diagnosis is where the deepest learning happens.

In your Claude conversation inside the Habit Forensics Lab project, paste the following 4-Law Forensic Pass prompt:

Coach, let's run the 4-Law Forensic Pass.

Before we start: ask me which of the Four Laws I think is most broken for my "ship daily" habit. I want to make my prediction before seeing the diagnosis. Record my prediction.

Then walk through each Law one at a time (approximately 4 minutes per Law):

**Law 1: Make It Obvious**
- Is my cue visible, specific, and unavoidable?
- Do I use an Implementation Intention? ("I will [BEHAVIOR] at [TIME] in [LOCATION]")
- Is there a Habit Stack anchoring this to an existing habit?
- Diagnose: Is this Law intact or broken? If broken, design a specific repair.

**Law 2: Make It Attractive**
- Is there a craving pull toward this habit when the cue fires?
- Could I use Temptation Bundling?
- Does my environment make this appealing or does it make competing behaviors more appealing?
- Diagnose: Intact or broken? Specific repair.

**Law 3: Make It Easy**
- What is the friction score? (How many steps between me and starting the habit?)
- Can I apply the Two-Minute Rule? What does the 2-minute version look like?
- What competing behaviors currently have lower friction?
- Diagnose: Intact or broken? Specific repair.

**Law 4: Make It Satisfying**
- Is there an immediate reward after I ship?
- Do I track completion visibly? (Habit Tracker, checkmark, streak counter?)
- Does "Never Miss Twice" apply? When I miss a day, do I recover the next day?
- Diagnose: Intact or broken? Specific repair.

After all 4 Laws, present:
1. Summary: which Laws are intact, which are broken, what is the dominant failure mode
2. Compare my prediction to the actual diagnosis. Highlight the delta.
3. One environment design move I should make in my physical workspace today

Score each Law 0-2 (0 = broken, 1 = partially intact, 2 = strong). Total possible: 8. My target is to get the total to 6+ after repairs.

What does this prompt do?

This prompt structures a 4-part diagnostic conversation. Coach will first ask for your prediction, then walk through each of the Four Laws one at a time, spending about 4 minutes on each.

For every Law, Coach asks you a diagnostic question, you answer from your real experience, and Coach decides if that Law is intact or broken. If it is broken, Coach designs a specific repair tailored to your life.

Press Enter to send the prompt.

Coach should immediately ask you which of the Four Laws you think is most broken. Take a moment to think about this before answering.

What are the Four Laws?

Each Law targets a different stage of the Habit Loop you scored in the previous step.

Law 1: Make It Obvious fixes the Cue. It uses Implementation Intentions ("I will [BEHAVIOR] at [TIME] in [LOCATION]") and Habit Stacking (anchoring new behavior to an existing routine).
Law 2: Make It Attractive fixes the Craving. It uses Temptation Bundling (pairing the habit with something you enjoy).
Law 3: Make It Easy fixes the Response. It uses the Two-Minute Rule (scale the habit down to under 2 minutes to start) and friction reduction.
Law 4: Make It Satisfying fixes the Reward. It uses visible tracking and the "Never Miss Twice" rule (missing once is an accident, missing twice starts a new pattern).

Type your prediction of which Law is most broken and press Enter.
Answer each of Coach's diagnostic questions one at a time, using specific examples from your real experience.
Work through all four Laws. Spend about 4 minutes on each.

Answer from your real life

The quality of Coach's diagnosis depends entirely on the honesty of your answers. If Coach pushes back and says your answer is too vague, give a concrete example from the last time this habit died. Vague input produces vague repairs.

Coach sounds generic or skips your prediction?

If Coach jumps straight into the Laws without asking for your prediction, type: "Stop. You skipped my prediction. Ask me which Law I think is most broken before we start." Coach should reset.

If Coach sounds sycophantic or overly agreeable, check that your Project Instructions still contain the full Coach system prompt. Open Project Settings via the gear icon and verify the text is there.

Help me get Coach back on track with the 4-Law diagnostic

Review the forensic findings

After walking through all four Laws, Coach presents a summary. This is the moment where the diagnostic comes together into a single picture.

Read Coach's summary of which Laws are intact, which are broken, and what the dominant failure mode is.
Compare Coach's diagnosis to your prediction from the start of this step. If your prediction was wrong, pay close attention to the delta.
Note Coach's environment design move recommendation. This is a physical change to your actual workspace that supports the repair.

Why does the prediction delta matter?

Most people blame willpower when a habit dies. The prediction exercise forces you to name a specific Law before seeing the evidence. When your guess is wrong, that gap reveals a blind spot in how you think about your own behavior.

The blind spot is usually the real problem. You cannot fix what you cannot see.

Coach scores each Law 0-2, giving you a total out of 8. Your target after implementing the repairs is 6 or higher.

Save your forensic findings

Your forensic findings need a permanent home in your Obsidian vault. You will also save a backup of the prompt itself so you can re-run this diagnostic on any future habit.

In Obsidian, create a new note called forensic-findings.md at the root of your habit-forensics vault (the same level as habit-autopsy.md).
Paste the following template and fill in each section with your actual findings from Coach:

# 4-Law Forensic Findings

## My Prediction
[Learner's predicted most-broken Law]

## Law 1 - Make It Obvious
- **Status**: [ ] Intact [ ] Broken
- **Score**: /2
- **Finding**: [Coach's diagnosis]
- **Repair**: [Specific repair]

## Law 2 - Make It Attractive
- **Status**: [ ] Intact [ ] Broken
- **Score**: /2
- **Finding**: [Coach's diagnosis]
- **Repair**: [Specific repair]

## Law 3 - Make It Easy
- **Status**: [ ] Intact [ ] Broken
- **Score**: /2
- **Finding**: [Coach's diagnosis]
- **Repair**: [Specific repair]

## Law 4 - Make It Satisfying
- **Status**: [ ] Intact [ ] Broken
- **Score**: /2
- **Finding**: [Coach's diagnosis]
- **Repair**: [Specific repair]

## Summary
**Dominant failure mode**: [Coach's identified dominant failure mode]
**Prediction vs. actual delta**: [What was different from prediction]
**Environment design move**: [One physical workspace change to make today]

What goes in each section?

My Prediction is the Law you guessed was most broken before Coach diagnosed it.
Finding is Coach's specific diagnosis for that Law (what is broken or working).
Repair is the concrete fix Coach designed for that Law (e.g., a specific Implementation Intention or Temptation Bundle).
Environment design move is the one physical change to your workspace Coach recommended.

Replace every bracketed placeholder with your real findings from the Claude conversation.
Check the Status checkbox for each Law (Intact or Broken) and fill in your scores.
Save forensic-findings.md.

Now save a backup of the prompt so you can re-run this diagnostic on any future habit.

Create a new note called forensic-pass.md inside the prompts folder of your vault.
Paste the following into it:

# 4-Law Forensic Pass Prompt

Paste this into your Claude conversation after completing the Habit Autopsy.

---

Coach, let's run the 4-Law Forensic Pass.

Before we start: ask me which of the Four Laws I think is most broken for my "ship daily" habit. I want to make my prediction before seeing the diagnosis. Record my prediction.

Then walk through each Law one at a time (approximately 4 minutes per Law):

**Law 1: Make It Obvious**
- Is my cue visible, specific, and unavoidable?
- Do I use an Implementation Intention? ("I will [BEHAVIOR] at [TIME] in [LOCATION]")
- Is there a Habit Stack anchoring this to an existing habit?
- Diagnose: Is this Law intact or broken? If broken, design a specific repair.

**Law 2: Make It Attractive**
- Is there a craving pull toward this habit when the cue fires?
- Could I use Temptation Bundling?
- Does my environment make this appealing or does it make competing behaviors more appealing?
- Diagnose: Intact or broken? Specific repair.

**Law 3: Make It Easy**
- What is the friction score? (How many steps between me and starting the habit?)
- Can I apply the Two-Minute Rule? What does the 2-minute version look like?
- What competing behaviors currently have lower friction?
- Diagnose: Intact or broken? Specific repair.

**Law 4: Make It Satisfying**
- Is there an immediate reward after I ship?
- Do I track completion visibly? (Habit Tracker, checkmark, streak counter?)
- Does "Never Miss Twice" apply? When I miss a day, do I recover the next day?
- Diagnose: Intact or broken? Specific repair.

After all 4 Laws, present:
1. Summary: which Laws are intact, which are broken, what is the dominant failure mode
2. Compare my prediction to the actual diagnosis. Highlight the delta.
3. One environment design move I should make in my physical workspace today

Score each Law 0-2 (0 = broken, 1 = partially intact, 2 = strong). Total possible: 8. My target is to get the total to 6+ after repairs.

Save prompts/forensic-pass.md.

✔️ Awesome, I've got everything!

Double check you have saved both forensic-findings.md and prompts/forensic-pass.md in your Obsidian vault.

ⓧ I'd like to double check the full notes

Your forensic-findings.md should have this structure (with your real findings filled in for the bracketed placeholders):

# 4-Law Forensic Findings

## My Prediction
[Learner's predicted most-broken Law]

## Law 1 - Make It Obvious
- **Status**: [ ] Intact [ ] Broken
- **Score**: /2
- **Finding**: [Coach's diagnosis]
- **Repair**: [Specific repair]

## Law 2 - Make It Attractive
- **Status**: [ ] Intact [ ] Broken
- **Score**: /2
- **Finding**: [Coach's diagnosis]
- **Repair**: [Specific repair]

## Law 3 - Make It Easy
- **Status**: [ ] Intact [ ] Broken
- **Score**: /2
- **Finding**: [Coach's diagnosis]
- **Repair**: [Specific repair]

## Law 4 - Make It Satisfying
- **Status**: [ ] Intact [ ] Broken
- **Score**: /2
- **Finding**: [Coach's diagnosis]
- **Repair**: [Specific repair]

## Summary
**Dominant failure mode**: [Coach's identified dominant failure mode]
**Prediction vs. actual delta**: [What was different from prediction]
**Environment design move**: [One physical workspace change to make today]

Your prompts/forensic-pass.md should match the full prompt you pasted into Claude earlier in this step, including the header and instructions line.

Not sure what to write in the findings?

Scroll back through your Claude conversation to find Coach's diagnosis for each Law. Copy the key sentences directly. You do not need to rewrite them in your own words.

If Coach did not clearly score a Law or skipped the environment design move, type: "Coach, you missed the score for Law [number]. Score it 0-2 and give me the environment design move."

Help me extract my forensic findings from Coach's responses

You now have a complete diagnosis of what is broken in your "ship daily" habit. Next up, you will turn these findings into a reusable 1-page Habit Operating System and write your Identity Vote.

Build Your Identity Reframe and Habit Operating System

You now know exactly which Laws of Behavior Change are broken for your daily shipping habit. You have scores, findings, and repairs.

Knowing which Laws are broken is only half the diagnostic. The other half is reframing your habit from an outcome ("ship 365 projects") into an identity ("become a person who ships daily"). In this step, you will write an Identity Vote sentence, compile everything into a reusable 1-page Habit Operating System, and close with a formal After-Action Review.

In this step, get ready to:

Write an Identity Vote sentence using Coach's Identity Future voice.
Build a 1-page Habit Operating System that works on any future failing habit.
Run a US Army-style After-Action Review on the entire forensic session.

Write your Identity Vote

Most people frame habits as outcomes: "I want to complete 365 projects." The problem is that outcomes do not change behavior. Identity changes behavior. James Clear puts it this way: "Every action is a vote for the type of person you wish to become."

Coach has a voice called Identity Future that will refuse outcome-only language and push you to write a sentence grounded in identity. Let's invoke it.

Create a new note in your Obsidian vault at prompts/identity-reframe.md and paste the following content:

# Identity Reframe Prompt

Paste this into your Claude conversation after completing the Forensic Pass.

---

Coach, let's do the Identity Reframe.

Invoke the Identity Future voice. I am going to try to write my Identity Vote sentence. Your job:

1. If I write an outcome-only sentence (e.g., "I want to complete 365 projects"), refuse it. Demand identity language.
2. If I use weak nouns (e.g., "someone who tries"), refuse it. Demand a specific, ownable noun.
3. If I write in future tense (e.g., "I will become..."), refuse it. Demand present tense.

The format I am aiming for:
"Every time I [specific action], I cast a vote for being a [identity noun] who [identity verb]."

Keep pushing until the sentence is:
- First person
- Present tense
- Names a specific noun (builder, creator, shipper, maker)
- Names a specific action (ship a project, push code, publish, deploy)
- Feels ownable, not performative

Once the Identity Vote is strong, return to Coach voice and confirm the sentence.

What does this prompt do?

This prompt activates Coach's Identity Future voice. Identity Future is strict. It will reject any sentence that frames your habit as an outcome ("I want to complete 365 projects") and demand identity language instead ("I am a builder who ships daily").

The goal is a single sentence in this format: "Every time I [specific action], I cast a vote for being a [identity noun] who [identity verb]." First-person, present-tense, with a concrete noun like "builder" or "shipper."

Save prompts/identity-reframe.md in Obsidian.
In your existing Claude conversation inside the Habit Forensics Lab project, copy everything below the --- line from the prompt file and paste it into the conversation.
Press Enter to send the prompt to Coach.

Coach will now ask you to write your first attempt at an Identity Vote sentence. Go ahead and type one.

Expect pushback

Your first draft will probably get rejected. If you write something like "I want to ship 365 projects this year," Identity Future will push back: "That is an outcome. Who is the person who does this? What noun are you?" This is by design. Keep refining until Coach confirms the sentence.

Keep responding to Coach until Identity Future accepts your Identity Vote sentence.
Verify that Coach returns to its normal voice and confirms the sentence after accepting it.

Your confirmed sentence should look something like: "Every time I ship a NextWork project, I cast a vote for being a builder who ships daily." The specific nouns and actions will be yours.

Coach not pushing back on weak sentences?

If Coach accepts vague or outcome-only language without challenging it, the Identity Future voice may not have activated properly. Try typing: "Invoke Identity Future and re-evaluate my last sentence. Is it truly identity language, or is it an outcome in disguise?"

If Coach still sounds generic, check that your Project Instructions in the project settings still contain the full Coach system prompt from Step 1. Help me get Coach's Identity Future voice working properly.

Build the 1-page Habit Operating System

You now have a complete set of findings: failure patterns from the Autopsy, Habit Loop scores, 4-Law diagnoses with repairs, and an Identity Vote. The Habit Operating System pulls all of this into a single reusable document.

The key word is reusable. When a future habit starts dying, you run the same diagnostic on the new habit and fill in a fresh OS. You will never need to invent the process from scratch again.

Create a new note in the root of your habit-forensics Obsidian vault called habit-os.md and paste the following template:

# Habit Operating System

*Reusable 1-page diagnostic. Fill in for any failing habit.*

---

## 1. Habit Statement

**Habit**: [What you do]
**Cue**: [When and where]
**Implementation Intention**: "I will [BEHAVIOR] at [TIME] in [LOCATION]"
**Habit Stack**: "After I [EXISTING HABIT], I will [THIS HABIT]"

## 2. Habit Loop Scorecard

| Stage | Score (0-2) | Notes |
|-------|-------------|-------|
| Cue | /2 | |
| Craving | /2 | |
| Response | /2 | |
| Reward | /2 | |
| **Total** | **/8** | |

## 3. 4-Law Diagnostic

**Law 1 - Make It Obvious**: [ ] Intact [ ] Broken
- Finding:
- Repair:

**Law 2 - Make It Attractive**: [ ] Intact [ ] Broken
- Finding:
- Repair:

**Law 3 - Make It Easy**: [ ] Intact [ ] Broken
- Finding:
- Repair:

**Law 4 - Make It Satisfying**: [ ] Intact [ ] Broken
- Finding:
- Repair:

**Dominant failure mode**:
**Environment design move**:

## 4. Identity Vote

> "Every time I _____________, I cast a vote for being a _____________ who _____________."

## 5. 30-Day Re-Score

**Date of original diagnostic**:
**Re-score date (30 days later)**:
**Movement criteria**:
- [ ] Habit frequency increased (from ___ days/week to ___ days/week)
- [ ] Identity language adopted in daily self-talk
- [ ] If neither: re-run 4-Law Forensic on lowest-scoring Law

What are the five sections?

Habit Statement captures the what, when, and where of your habit, plus your Implementation Intention and Habit Stack.
Habit Loop Scorecard records your Cue, Craving, Response, and Reward scores from the Autopsy.
4-Law Diagnostic logs which Laws are broken and the specific repair for each.
Identity Vote holds the first-person sentence you just wrote with Coach.
30-Day Re-Score defines what "movement" looks like a month from now so you can objectively measure progress.

Open your habit-autopsy.md and forensic-findings.md notes side by side in Obsidian.
Fill in Section 1 (Habit Statement) using your declared habit from the Autopsy. Include a specific Implementation Intention with a real time and location from your life.
Fill in Section 2 (Habit Loop Scorecard) by copying the scores from your habit-autopsy.md scorecard.
Fill in Section 3 (4-Law Diagnostic) using the findings and repairs from forensic-findings.md. Mark each Law as Intact or Broken.
Fill in Section 4 (Identity Vote) with the confirmed sentence from your Claude conversation.
Fill in Section 5 (30-Day Re-Score) with today's date and a date 30 days from now. Fill in your current habit frequency so you can compare later.
Save habit-os.md.

Switch to Reading View in Obsidian by clicking the book icon in the top-right corner, or pressing Ctrl+E. You should see a cleanly formatted document with all five sections filled in, including a populated table for your Habit Loop Scorecard.

Table not rendering properly?

Markdown tables require exact spacing. Make sure each row has the same number of pipe characters (|) as the header row. The separator row (|-------|) must also match. If a cell looks broken, check for missing or extra pipe characters on that line. Help me fix my markdown table formatting in Obsidian.

Run the After-Action Review

The US Army developed the After-Action Review (AAR) as a structured debrief format. It forces you to compare what you intended to happen with what actually happened, then name specific next actions. No vague "I'll try harder" allowed.

Create a new note in your Obsidian vault at prompts/aar.md and paste the following content:

# After-Action Review Prompt

Paste this into your Claude conversation after building the Habit OS.

---

Coach, let's run the After-Action Review. Use the US Army 4-question AAR format (from TC 25-20):

1. **What did we intend to happen?** (What was the plan going into this forensic session?)
2. **What actually happened?** (What did we actually discover? What surprised us?)
3. **Why was there a difference?** (Why did my prediction not match the diagnosis? Why had I not seen the pattern before?)
4. **What will we do differently next time?** (What specific actions will I take in the next 7 days based on this diagnostic?)

Ask me each question one at a time. Do not accept vague answers. If I say "I'll try harder," push back: "That is not an action. Name the specific behavior, time, and location."

What does this prompt do?

The AAR format comes from US Army Training Circular TC 25-20. It has four questions: What did we intend? What actually happened? Why the difference? What will we do differently?

Coach is instructed to ask these one at a time and reject vague answers. If you say "I'll try harder," Coach will push back and demand a specific behavior, time, and location.

Save prompts/aar.md in Obsidian.
In your Claude conversation, copy everything below the --- line from the prompt file and paste it into the conversation.
Press Enter to send.

Coach will ask you each of the four AAR questions one at a time. Answer each honestly based on what you planned coming into this session versus what the diagnostic actually uncovered.

Focus on the prediction delta

Question 3 ("Why was there a difference?") is where the deepest learning happens. Think about why your prediction of the most broken Law did not match the actual diagnosis. That gap is the blind spot your system needs to account for.

Answer all four AAR questions in your Claude conversation.
Verify that Coach accepted all four answers. If Coach pushed back on any answer as too vague, revise and resubmit until it accepts.

Now save your AAR as a permanent record in Obsidian.

Create a new note at the root of your habit-forensics vault called aar.md and paste the following template:

# After-Action Review

## 1. What did we intend to happen?
[Learner's answer about the plan going into the forensic session]

## 2. What actually happened?
[Learner's answer about discoveries and surprises]

## 3. Why was there a difference?
[Learner's answer about why prediction didn't match diagnosis]

## 4. What will we do differently next time?
[Learner's specific actions with behavior, time, and location named]

What does each section capture?

Section 1 records what you expected to find when you started the forensic session.
Section 2 records the actual findings and any surprises from the diagnostic.
Section 3 captures why your prediction was wrong. This is where you name the blind spot.
Section 4 names specific actions for the next 7 days, with behavior, time, and location. No vague intentions.

Replace each bracketed placeholder with your actual answers from the Claude conversation.
Save aar.md.
Switch to Reading View in Obsidian to confirm all four sections are filled in with specific, concrete answers.

Coach accepting vague answers?

If Coach did not push back on answers like "I'll try harder" or "I'll be more consistent," those are not actionable. Go back to your Claude conversation and type: "Coach, re-evaluate my answer to question 4. Is it specific enough? Does it name a behavior, a time, and a location?" Replace any vague answers in aar.md with the revised versions. Help me write more specific AAR answers.

✔️ Awesome, I've got everything!

Double check that all four new files are saved in your Obsidian vault: prompts/identity-reframe.md, prompts/aar.md, habit-os.md, and aar.md.

ⓧ I'd like to double check the full code

Here is the exact content for each file created in this step. Compare yours against these.

# Identity Reframe Prompt

Paste this into your Claude conversation after completing the Forensic Pass.

---

Coach, let's do the Identity Reframe.

Invoke the Identity Future voice. I am going to try to write my Identity Vote sentence. Your job:

1. If I write an outcome-only sentence (e.g., "I want to complete 365 projects"), refuse it. Demand identity language.
2. If I use weak nouns (e.g., "someone who tries"), refuse it. Demand a specific, ownable noun.
3. If I write in future tense (e.g., "I will become..."), refuse it. Demand present tense.

The format I am aiming for:
"Every time I [specific action], I cast a vote for being a [identity noun] who [identity verb]."

Keep pushing until the sentence is:
- First person
- Present tense
- Names a specific noun (builder, creator, shipper, maker)
- Names a specific action (ship a project, push code, publish, deploy)
- Feels ownable, not performative

Once the Identity Vote is strong, return to Coach voice and confirm the sentence.

# After-Action Review Prompt

Paste this into your Claude conversation after building the Habit OS.

---

Coach, let's run the After-Action Review. Use the US Army 4-question AAR format (from TC 25-20):

1. **What did we intend to happen?** (What was the plan going into this forensic session?)
2. **What actually happened?** (What did we actually discover? What surprised us?)
3. **Why was there a difference?** (Why did my prediction not match the diagnosis? Why had I not seen the pattern before?)
4. **What will we do differently next time?** (What specific actions will I take in the next 7 days based on this diagnostic?)

Ask me each question one at a time. Do not accept vague answers. If I say "I'll try harder," push back: "That is not an action. Name the specific behavior, time, and location."

# Habit Operating System

*Reusable 1-page diagnostic. Fill in for any failing habit.*

---

## 1. Habit Statement

**Habit**: [What you do]
**Cue**: [When and where]
**Implementation Intention**: "I will [BEHAVIOR] at [TIME] in [LOCATION]"
**Habit Stack**: "After I [EXISTING HABIT], I will [THIS HABIT]"

## 2. Habit Loop Scorecard

| Stage | Score (0-2) | Notes |
|-------|-------------|-------|
| Cue | /2 | |
| Craving | /2 | |
| Response | /2 | |
| Reward | /2 | |
| **Total** | **/8** | |

## 3. 4-Law Diagnostic

**Law 1 - Make It Obvious**: [ ] Intact [ ] Broken
- Finding:
- Repair:

**Law 2 - Make It Attractive**: [ ] Intact [ ] Broken
- Finding:
- Repair:

**Law 3 - Make It Easy**: [ ] Intact [ ] Broken
- Finding:
- Repair:

**Law 4 - Make It Satisfying**: [ ] Intact [ ] Broken
- Finding:
- Repair:

**Dominant failure mode**:
**Environment design move**:

## 4. Identity Vote

> "Every time I _____________, I cast a vote for being a _____________ who _____________."

## 5. 30-Day Re-Score

**Date of original diagnostic**:
**Re-score date (30 days later)**:
**Movement criteria**:
- [ ] Habit frequency increased (from ___ days/week to ___ days/week)
- [ ] Identity language adopted in daily self-talk
- [ ] If neither: re-run 4-Law Forensic on lowest-scoring Law

# After-Action Review

## 1. What did we intend to happen?
[Learner's answer about the plan going into the forensic session]

## 2. What actually happened?
[Learner's answer about discoveries and surprises]

## 3. Why was there a difference?
[Learner's answer about why prediction didn't match diagnosis]

## 4. What will we do differently next time?
[Learner's specific actions with behavior, time, and location named]

You now have a complete diagnostic system: an Identity Vote, a filled-in Habit Operating System, and a formal After-Action Review. Next up, you will build two visual artifacts in Obsidian and set a 30-day calendar event to re-score your progress.

Create Your Visual Artifacts and Set the 30-Day Re-Score

You now have a completed Habit Operating System, a full After-Action Review, and an Identity Vote sentence that reframes your daily shipping habit as identity. The diagnostic work is done.

In this step, you will turn your findings into two visual artifacts that live inside your Obsidian vault. Then you will set a calendar reminder to re-score your habit in 30 days so the diagnostic does not just sit in a folder collecting dust.

In this step, get ready to:

Build a Habit Loop diagnostic flowchart using Mermaid in Obsidian.
Build a 1% compounding curve chart showing how small daily gains multiply over a year.
Set a 30-day re-score calendar event to check whether your repairs are working.

Build the Habit Loop diagnostic flowchart

Mermaid is a diagram language that renders charts from plain text code blocks. Obsidian supports Mermaid natively, so you can write a diagram in markdown and see it render without any plugins.

In Obsidian, open your habit-forensics vault and navigate to the diagrams folder in the left sidebar.
Create a new note inside diagrams called habit-loop.md.
Paste the following content into the note:

# Habit Loop Diagnostic

```mermaid
flowchart LR
    CUE["1. CUE\nScore: _/2\nYour trigger fires"]
    CRAVING["2. CRAVING\nScore: _/2\nYou want to respond"]
    RESPONSE["3. RESPONSE\nScore: _/2\nYou do the habit"]
    REWARD["4. REWARD\nScore: _/2\nYou feel satisfied"]

    CUE --> CRAVING
    CRAVING --> RESPONSE
    RESPONSE --> REWARD
    REWARD -->|"Loop repeats"| CUE

    style CUE fill:#4A90D9,stroke:#333,color:#fff
    style CRAVING fill:#F5A623,stroke:#333,color:#fff
    style RESPONSE fill:#D0021B,stroke:#333,color:#fff
    style REWARD fill:#7ED321,stroke:#333,color:#fff
```

**Instructions**: Replace the `_/2` placeholders with your actual scores from the Habit Autopsy. Switch to Reading View (Ctrl+E) to see the diagram.

What does this diagram do?

This is a left-to-right flowchart (flowchart LR) that maps the four stages of the Habit Loop as a circular flow. The arrow from REWARD back to CUE creates the loop.

Each node contains a placeholder score (_/2) that you will replace with your actual scores from habit-autopsy.md. The style lines at the bottom color-code each stage so the diagram is easy to scan at a glance.

Switch to Reading View by clicking the book icon in the top-right corner of the note, or press Ctrl+E (Windows) or Cmd+E (macOS).

You should see a colored flowchart with four boxes (blue CUE, orange CRAVING, red RESPONSE, green REWARD) connected by arrows, with the last arrow looping back from REWARD to CUE.

Switch back to editing mode by pressing Ctrl+E (Windows) or Cmd+E (macOS).
Open your habit-autopsy.md note and find your Habit Loop Scorecard scores.
Back in diagrams/habit-loop.md, replace each _/2 placeholder with your actual score for that stage (e.g. change Score: _/2 to Score: 1/2).
Switch to Reading View again to confirm your personalized scores appear in each node.

Side-by-side editing

If you want to see the code and the rendered diagram at the same time, right-click the tab and select Split right. Set one pane to Source mode and the other to Reading mode.

Diagram not rendering?

Make sure the code fence starts with exactly ```mermaid (three backticks, no spaces before mermaid). Check that you are in Reading View, not editing mode.

If you see raw text instead of a diagram, try closing and reopening the note.

Why isn't my Mermaid diagram rendering in Obsidian?

Build the 1% compounding curve

One of the most powerful ideas in Atomic Habits is that 1% daily improvement does not add up linearly. It compounds exponentially. This chart makes that invisible math visible.

Create a new note in your diagrams folder called compounding-curve.md.
Paste the following content into the note:

# The 1% Compounding Curve

```mermaid
xychart
    title "1% Better Every Day for a Year (1.01^n)"
    x-axis "Day" [0, 30, 60, 90, 120, 150, 180, 210, 240, 270, 300, 330, 365]
    y-axis "Multiplier" 0 --> 40
    line [1, 1.35, 1.82, 2.45, 3.3, 4.45, 6, 8.1, 10.94, 14.74, 19.89, 26.83, 37.78]
```

**What this shows**: If you improve just 1% every day, you do not grow linearly. You grow exponentially. By day 365, you are 37.78 times where you started. The math: 1.01 raised to the power of 365 equals 37.78.

**Important caveat**: This is motivational framing from Atomic Habits, not literal measurement. You cannot measure a literal 1% improvement in most life domains. The point is that tiny consistent actions compound into massive results over time. The curve makes the invisible visible.

**If this diagram does not render**: Your Obsidian version may not support the xychart diagram type. Copy the code block and paste it into the Mermaid Live Editor at https://mermaid.live to see it render. Take a screenshot and paste it back into this note.

What does this chart show?

The xychart diagram type plots a line chart with an x-axis (days 0 through 365) and a y-axis (the multiplier). Each data point is the result of 1.01 raised to the power of that day number.

The curve is flat for the first few months. Then it accelerates sharply. That shape is the whole point. Consistency feels invisible early on, but it is building something massive underneath.

Switch to Reading View by pressing Ctrl+E (Windows) or Cmd+E (macOS) to see the chart render.

You should see a line chart that starts nearly flat on the left and curves steeply upward on the right, reaching 37.78 by day 365.

Seeing an error instead of a chart?

The xychart diagram type requires Mermaid v10.3.0 or later. Some older versions of Obsidian may not support it.

If the chart does not render, copy everything inside the ```mermaid code fence (starting from xychart through the last data point). Open mermaid.live in your browser, paste the code into the editor on the left, and the chart renders on the right. Take a screenshot and paste it into your note.

My Mermaid xychart is not rendering in Obsidian. How do I fix this?

✔️ Awesome, both diagrams are rendering

Your visual artifacts are done. Make sure both notes are saved in your diagrams folder.

ⓧ I'd like to double check the full files

Here is the full content for diagrams/habit-loop.md:

# Habit Loop Diagnostic

```mermaid
flowchart LR
    CUE["1. CUE\nScore: _/2\nYour trigger fires"]
    CRAVING["2. CRAVING\nScore: _/2\nYou want to respond"]
    RESPONSE["3. RESPONSE\nScore: _/2\nYou do the habit"]
    REWARD["4. REWARD\nScore: _/2\nYou feel satisfied"]

    CUE --> CRAVING
    CRAVING --> RESPONSE
    RESPONSE --> REWARD
    REWARD -->|"Loop repeats"| CUE

    style CUE fill:#4A90D9,stroke:#333,color:#fff
    style CRAVING fill:#F5A623,stroke:#333,color:#fff
    style RESPONSE fill:#D0021B,stroke:#333,color:#fff
    style REWARD fill:#7ED321,stroke:#333,color:#fff
```

**Instructions**: Replace the `_/2` placeholders with your actual scores from the Habit Autopsy. Switch to Reading View (Ctrl+E) to see the diagram.

And here is the full content for diagrams/compounding-curve.md:

# The 1% Compounding Curve

```mermaid
xychart
    title "1% Better Every Day for a Year (1.01^n)"
    x-axis "Day" [0, 30, 60, 90, 120, 150, 180, 210, 240, 270, 300, 330, 365]
    y-axis "Multiplier" 0 --> 40
    line [1, 1.35, 1.82, 2.45, 3.3, 4.45, 6, 8.1, 10.94, 14.74, 19.89, 26.83, 37.78]
```

**What this shows**: If you improve just 1% every day, you do not grow linearly. You grow exponentially. By day 365, you are 37.78 times where you started. The math: 1.01 raised to the power of 365 equals 37.78.

**Important caveat**: This is motivational framing from Atomic Habits, not literal measurement. You cannot measure a literal 1% improvement in most life domains. The point is that tiny consistent actions compound into massive results over time. The curve makes the invisible visible.

**If this diagram does not render**: Your Obsidian version may not support the xychart diagram type. Copy the code block and paste it into the Mermaid Live Editor at https://mermaid.live to see it render. Take a screenshot and paste it back into this note.

Set your 30-day re-score calendar event

A diagnostic is only useful if you check whether the repairs actually worked. Without a re-score date, this Habit OS becomes another abandoned document.

Open your calendar app (Windows Calendar, Google Calendar, Outlook, or whichever calendar you use daily).
Create a new event 30 days from today.
Set the title to:

Habit Forensics Re-Score: Ship Daily

In the event description field, paste the following re-score instructions:

Re-run the Habit Loop Scorecard (Cue/Craving/Response/Reward, 0-2 each). Compare to original scores. Movement criteria: habit frequency increased OR identity language adopted in daily self-talk. If neither, re-run the 4-Law Forensic on the lowest-scoring Law.

Save the calendar event.

Why 30 days?

Thirty days is enough time for the repairs to take effect without being so long that you forget the diagnostic entirely. The re-score is not pass/fail. You are looking for movement in one of two directions: your habit frequency went up, or you started using identity language in your self-talk.

If neither happened, the description tells you exactly what to do. Re-run the 4-Law Forensic on whichever Law scored lowest. The system repairs itself.

Not sure which calendar app to use?

Use whichever calendar you actually check every day. The best reminder system is the one you already look at. If you do not use a calendar app regularly, set a phone alarm or reminder instead.

How do I set a reliable 30-day reminder for my habit re-score?

Secret mission

The Teach-Back

You have run the prompts and built the documents. But can you teach the framework from memory? Record a 3-minute Teach-Back explaining the Four Laws and identity framework to someone who has never read Atomic Habits.

Before you go

Clean Up Your Resources

Decide whether to keep your resources running, pause them to come back later, or delete them entirely. This project runs entirely locally using Claude Desktop and Obsidian, so there are no ongoing cloud costs beyond your existing Claude Pro subscription.

Resources you used:

Claude Project: Habit Forensics Lab (with Coach persona loaded in Project Instructions).
Obsidian vault: habit-forensics (containing prompts, diagrams, Habit OS, AAR, and autopsy notes).
Calendar event: Habit Forensics Re-Score: Ship Daily (30 days from your project date).

Keep everything running

No action needed. Choose this if you plan to reuse the Coach persona on other habits or want to revisit your diagnostic findings.

Your Claude Project stays in the sidebar, ready for new conversations with Coach loaded.
Your Obsidian vault stays on disk with all notes and diagrams intact.
Your 30-day calendar event stays scheduled to remind you to re-score.

Pause - I'll come back to this later

Keep your files but tidy up active resources so they are not cluttering your workspace.

In Claude Desktop, you can leave the Habit Forensics Lab project as-is. It does not consume resources when you are not chatting in it.
Close the habit-forensics vault in Obsidian by clicking Open another vault in the bottom-left corner. The vault folder stays on your computer for when you return.
Keep the 30-day calendar event active so you remember to come back and re-score.

Delete - I don't want to use this again

Remove all project resources and start fresh.

In Claude Desktop, open the Habit Forensics Lab project. Click the gear icon to open Project Settings, then scroll down and click Delete Project.
Delete the habit-forensics folder from your computer. On macOS, drag it to the Trash. On Windows, right-click and select Delete.
Open your calendar app and delete the Habit Forensics Re-Score: Ship Daily event.

Mission accomplished

That's a Wrap!

Nice work! You just built an AI Forensics Lab that you can point at any failing habit for the rest of your life.

The most valuable thing you built today is not the documents. It is the diagnostic skill. Next time any habit starts dying, you know exactly which questions to ask.

What you learned:

Built a reusable Coach persona in a Claude Desktop Project with two invokable voices (Past-Self-as-Pattern and Identity Future), then used it to diagnose a real failing habit through a Habit Autopsy, Habit Loop Scorecard, and a full 4-Law Forensic Pass.
Wrote an Identity Vote sentence that reframes your habit as identity instead of outcome, and assembled a 1-page Habit Operating System that works on any future habit.
Created two Mermaid visual artifacts (a Habit Loop flowchart and a 1% compounding curve) rendering live in Obsidian, and set a 30-day calendar event to re-score your repairs.
Secret Mission: Recorded a 3-minute Teach-Back proving you can explain the Four Laws and identity framework to someone who has never read Atomic Habits.

Ready to quiz yourself?