WhyStrohm Content Infrastructure Audit

A two-phase content diagnostic: score your content against a 5-layer framework, then watch one piece get fixed live.

Flow

digraph audit_flow {
    "User runs /whystrohm-audit" [shape=doublecircle];
    "Ask for URL" [shape=box];
    "Scrape website" [shape=box];
    "Q1: Ideal buyer?" [shape=box];
    "Q2: Differentiator?" [shape=box];
    "Q3: Paste content sample" [shape=box];
    "Build voice profile" [shape=box];
    "Score against 5 layers" [shape=box];
    "Display score (number first)" [shape=box];
    "Show layer breakdown" [shape=box];
    "Show top 3 gaps" [shape=box];
    "Pick weakest piece" [shape=box];
    "Rewrite using framework" [shape=box];
    "Show before/after" [shape=box];
    "Reveal framework" [shape=box];
    "Show CTA" [shape=doublecircle];

    "User runs /whystrohm-audit" -> "Ask for URL";
    "Ask for URL" -> "Scrape website";
    "Scrape website" -> "Q1: Ideal buyer?";
    "Q1: Ideal buyer?" -> "Q2: Differentiator?";
    "Q2: Differentiator?" -> "Q3: Paste content sample";
    "Q3: Paste content sample" -> "Build voice profile";
    "Build voice profile" -> "Score against 5 layers";
    "Score against 5 layers" -> "Display score (number first)";
    "Display score (number first)" -> "Show layer breakdown";
    "Show layer breakdown" -> "Show top 3 gaps";
    "Show top 3 gaps" -> "Pick weakest piece";
    "Pick weakest piece" -> "Rewrite using framework";
    "Rewrite using framework" -> "Show before/after";
    "Show before/after" -> "Reveal framework";
    "Reveal framework" -> "Show CTA";
}

Phase 1: The Diagnostic

Step 1: Get the URL

Ask: "What's your website URL?"

Nothing else. One question. Wait for answer.

Step 2: Scrape the Website

Use WebFetch to pull:

Homepage
About or Services page (look for /about, /services, /what-we-do, or similar)
Most recent blog post (if /blog exists)

While scraping, tell the user: "Pulling your site now — analyzing voice, positioning, and content patterns..."

Step 3: Build Voice Profile

Read rules/voice-analysis.md. Build the internal voice profile from the scraped pages. Do NOT display this to the user — it's used for scoring.

Step 4: Ask Three Questions

Ask these ONE AT A TIME. Wait for each answer before asking the next.

"Who is your ideal buyer?" (Role, company size, industry)
"What's the one thing you do that nobody else does the same way?"
"Paste one piece of content you've published recently — a LinkedIn post, email, blog paragraph, anything."

Step 5: Score the Content

Read rules/scoring-rubric.md. Score the pasted content sample (Q3) against all 5 layers. Also cross-reference with scraped website content for voice consistency.

For each layer:

Assign a score (1-10)
Pull exact quotes as evidence
Note specific violations

Step 6: Display the Report

Read templates/audit-report.md. Follow the format exactly.

Critical: Show the score number FIRST. {score}/50 displayed prominently. Let it land. THEN show the layer breakdown, THEN the top 3 gaps, THEN the framework explanation.

Do not rush past the score. The number is the emotional hook.

Phase 2: The Live Fix

Step 7: Pick the Weakest Piece

Identify the lowest-scoring content (the pasted sample or a scraped page section). Tell the user:

"This scored {X}/50. Let me show you what it looks like through the system."

Step 8: Rewrite It

Read rules/rewrite-rules.md. Apply all 5 layers to the weakest piece:

Fix vocabulary (Layer 1)
Add structural elements (Layer 2)
Insert proof (Layer 3)
Align voice (Layer 4)
Flip to buyer perspective (Layer 5)

Step 9: Before/After

Display the original and rewritten version side by side using the format from templates/audit-report.md. Show per-layer scores on both with deltas.

Step 10: Framework Reveal + CTA

Show the framework explanation from templates/audit-report.md (the "System Behind This Score" section).

Then read templates/cta.md and display the closing pitch.

Rules

One question at a time. Never batch questions.
Score first, explain second. Always.
Quote their content. Never paraphrase when showing problems.
No emojis. Ever.
No hype in the audit itself. The skill must practice what it preaches — proof-dense, specific, zero hollow language.
Don't apologize or soften. The audit is honest. "Your content scored 14/50" not "Your content has some room for improvement."

Related Skills

Digital Twin — Extract your voice into a reusable AI System Prompt before auditing. Includes a 10-dimension scoring rubric and 15 stress tests for validating the extraction.
Voice Scorer (/whystrohm-voice-scorer) — Measure drift between your website voice and social content. Runs one layer of what this audit covers across all five.

whystrohm-audit