Voice Interview

Instructions

When the user runs /voice:

Step 0: Load Configuration

Find the .amplifyrc config file to resolve the vault path and voice file location. Check these locations in order:

1. .amplifyrc in the current working directory
2. .amplifyrc at the path specified by VAULT_PATH environment variable
3. .amplifyrc at the fallback vault path (if configured in CLAUDE.md or known from previous sessions)

Read the config and extract:

vaultPath    → absolute path to the Obsidian vault
paths.voice  → voice file relative to vault root (default: "Amplify/Voice.md")

Step 1: Check for Existing Voice.md

Read the existing Voice.md at {vaultPath}/{paths.voice}.

• If it exists and has real content (not just placeholder brackets), ask: "You already have a Voice.md. Want to start fresh or update specific sections?"
• If it exists but is the blank template (has [Your name or [What you placeholder text), proceed to the interview.
• If it doesn't exist, proceed to the interview.

Step 2: Ask the Five Questions

Use AskUserQuestion or natural conversation to ask these five questions one at a time. Wait for each answer before asking the next.

Question 1: Who I am "What's your name, what do you do, and what's your background? Tell me whatever feels relevant."

Question 2: My mission "What are you trying to accomplish? What change do you want to make?"

Question 3: My audience "Who are you talking to? What do they need? What keeps them up at night?"

Question 4: What I believe "What are your core beliefs? What hills will you die on? What makes your perspective different?"

Question 5: How I talk "How do you talk? Direct? Warm? Technical? Casual? What words do you use or avoid?"

For each question:

• If the user gives a real answer, use it for that section.
• If the user says "skip", presses enter with no input, or says they'll do it later, use the placeholder text for that section.

Step 3: Build Voice.md

Construct the Voice.md content using the user's answers. Write in first person, preserving their words and voice. Clean up grammar if needed but don't rewrite their personality.

Placeholder text for skipped sections:

Format:

# Voice

## Who I am
{answer or placeholder}

## My mission
{answer or placeholder}

## My audience
{answer or placeholder}

## What I believe
{answer or placeholder}

## How I talk
{answer or placeholder}

Step 4: Write and Report

Write the file to {vaultPath}/{paths.voice}.

Report back:

Voice.md written to {paths.voice}

Sections filled: {count}/5
{if any skipped: "Run /voice again anytime to fill in the rest."}

Humanizer Pass

Before writing, run a humanizer pass on any prose you wrote (not the user's own words). No AI vocabulary, no filler, no significance inflation. The user's voice should come through, not yours.