Blog Rewriter: Optimize Existing Posts

Rewrites and optimizes existing blog posts for dual ranking: Google search and AI citation platforms. Preserves the author's voice while applying the 6 pillars of optimization.

Key references:

• references/quality-scoring.md - 5-category scoring (Content 30, SEO 25, E-E-A-T 15, Technical 15, AI Citation 15)
• references/eeat-signals.md - Experience, expertise, authority, trust markers
• references/internal-linking.md - Linking strategy and anchor text rules
• references/visual-media.md - Image sourcing and chart styling
• skills/blog/references/synthesis-contract.md - 6 LAWs for re-citation hygiene during rewrite (v1.8.0; cross-skill ref lives in the orchestrator's references dir)
• skills/blog/references/research-quality.md - cross-source clustering for replacement-statistic research (v1.8.0)

Cross-reference

For 21 evidence-led optimization prompts (AI-detector test, CTR audit, schema, PAA rewording, technical audit, ChatGPT visibility) directly applicable to rewrite work, see /blog flow optimize.

Workflow

Phase 1: Audit (Read-Only)

1. Read the blog post - Detect format (MDX, markdown, HTML)
2. Run the quality checklist against references/quality-scoring.md:
   • Count fabricated vs sourced statistics
   • Check answer-first formatting (H2 -> stat in first sentence?)
   • Count images and charts (type diversity?)
   • Measure paragraph lengths (any > 150 words?)
   • Check heading hierarchy (H1 -> H2 -> H3, no skips?)
   • Look for FAQ schema
   • Check freshness signals (lastUpdated, dateModified)
   • Assess self-promotion level
   • Evaluate citation tier quality
3. AI content detection scan:
   • Burstiness score - Measure sentence length variance across the post. Low variance (most sentences within 3-5 words of each other) is a strong AI signal. Calculate: standard deviation of sentence word counts. Target SD > 6.
   • Known AI phrase scan - Check for these high-frequency AI phrases:
     • "in today's digital landscape", "it's important to note", "dive into"
     • "game-changer", "navigate the landscape", "revolutionize", "seamlessly"
     • "cutting-edge", "harness the power of", "leverage" (as verb)
     • "delve", "crucial", "elevate", "foster", "landscape" (overused)
     • "multifaceted", "robust", "tapestry", "embark"
     • Full list in agents/blog-writer.md
   • Vocabulary diversity - Calculate Type-Token Ratio (TTR): unique words / total words. Low TTR (< 0.40) suggests AI-generated repetitive phrasing. Target TTR > 0.50 for natural prose.
   • AI content percentage estimate - Based on burstiness, phrase density, and TTR, estimate what percentage of the content reads as AI-generated (0-100%). Report as: "AI content estimate: ~X%"
   • Second-order structural reflex scan (v1.8.0) - The first-order checks above are vocabulary-level. The second-order pass catches what survives them: structural and rhythmic tics LLMs default to after the obvious words are replaced. Run against skills/blog/references/ai-slop-detection.md. Flag at minimum:
     • Question-cadence H2s above 70% of headings
     • Three or more "Here..." paragraph openers
     • Three-clause sentence rhythm above 50% in any 200-word window
     • More than 2 hedge words ("may," "often," "typically," "generally") in any 20-word span
     • Symmetric-list bloat (list-item word-count SD below 5)
     • More than 2 wrap-up rhetorical questions ("What does this mean for...?")
     • More than half of H2 openers starting with a transition word
     • "The key insight is..." or "What's important here is..." as sentence openers
     • Listicle pre-list intro above 250 words
     • Opening-word repetition: top three first-words above 25% share
     • Paragraph-shape SD below 25 (visual monotony) A draft is only "AI-detection clean" when both passes are clean. The two-namespace terminology (first-order/second-order for slop-detection vs Tier 1/2/3 for source authority) is intentional: see skills/blog/references/ai-slop-detection.md for why the labels diverged in v1.8.1.
4. Video embed check:
   • Count existing YouTube embeds in the post
   • If 0 embeds, flag: "No video embeds. YouTube has the strongest AI visibility correlation (0.737)"
   • If present, check: lazy loading? aria-labels? noscript fallback? VideoObject schema?
5. Cannibalization check:
   • Identify the post's primary keyword from title, H1, and first paragraph
   • Search the blog directory for other posts targeting the same keyword:
     • Grep headings and meta descriptions across all blog posts
     • Flag any posts with significant keyword overlap
   • If cannibalization found, report:
     • Which posts compete for the same keyword
     • Recommend: merge (combine into one stronger post) or differentiate (shift one post to a related but distinct keyword)
6. Calculate current score across 5 categories:
   • Score across 5 categories (Content Quality 30, SEO Optimization 25, E-E-A-T Signals 15, Technical Elements 15, AI Citation Readiness 15)
   • Total: 0-100
7. Present audit summary with specific findings, AI detection results, video status, cannibalization status, and score
8. Enter plan mode - Present section-by-section optimization plan

Wait for user approval before proceeding.

Phase 2: Research

1. Identify the blog's core topic from existing content
2. Find replacement statistics for any fabricated/unsourced data:
   • Search: [topic] study 2025 2026 data statistics
   • Target tier 1-3 sources only
3. Find images if post has fewer than 3:
   • Pixabay: site:pixabay.com [topic keywords]
   • Unsplash: site:unsplash.com [topic keywords]
   • Verify each URL returns HTTP 200
   • If nanobanana-mcp is configured, offer AI generation for missing/insufficient images via blog-image
4. Plan charts if post has fewer than 2:
   • Identify data suitable for visualization
   • Select diverse chart types

Phase 3: Chart Generation (Built-In)

When the post needs more visual elements, invoke the blog-chart sub-skill:

1. Select chart type using the diversity rule (no repeated types per post)
2. Pass: chart type, title, data values, source, platform format
3. Embed the returned SVG directly within a <figure> wrapper
4. Target 2-4 charts per 2,000-word post

See references/visual-media.md for chart type selection and styling rules.

Phase 4: Content Rewrite

Apply changes in this order:

4a. Preserve What Works

• Keep the author's voice and unique perspective
• Preserve original insights and first-hand experience
• Keep existing quality images and charts
• Maintain internal links

4b. Fix Frontmatter

• Add lastUpdated: "YYYY-MM-DD" (today's date)
• Keep original date unchanged
• Fix meta description: fact-dense, 150-160 chars, includes 1 statistic
• Add coverImage + coverImageAlt + ogImage if missing
  • Search Pixabay/Unsplash/Pexels for wide hero image (1200x630)
  • Or generate custom SVG cover via blog-chart (text-on-gradient with key stat)
  • Or generate custom AI image via blog-image sub-skill (if nanobanana-mcp configured)
• Verify tags/categories are appropriate

4c. Apply Answer-First Formatting

Every H2 section MUST open with a 40-60 word paragraph containing:

• At least one specific statistic with source attribution
• A direct answer to the heading's implicit question

4d. Replace Fabricated Statistics

• Search for patterns: "X% of...", "X out of Y...", unsourced claims
• Replace with real data from tier 1-3 sources
• Always include inline attribution: ([Source Name](url), year)

4e. Improve Headings

• Convert statement headings to questions where natural (60-70% target)
• Keep 2-3 statement headings for variety
• Ensure keyword appears in 2-3 headings naturally

4f. Fix Paragraph Length

• Split any paragrap