# Content Refresh > Re-optimize an existing content piece with current statistics, new sources, and fixed citations at light (~20%), medium (~50%), or heavy (~80%) rewrite scope — preserving evergreen sections, SEO keyword placements, the URL slug, and internal links, then re-running the pipeline's quality gates (fact-check, validation, humanizer, reviewer) and saving as a new version with a before/after comparison report; the original file is never overwritten. Triggers on "/contentforge:content-refresh", "update this old article", "rankings dropped, refresh this post", "this content has outdated stats", "refresh it with current data". Reuses the 10-phase pipeline agents and loads the original .docx from the brand's tracking backend; pairs with /contentforge:cf-audit to find candidates and /contentforge:publish to push the update. - Skill: `teachskillofskills-ai/content-refresh` (Agent Skill) - Install (CLI): `npx skillmds@latest add teachskillofskills-ai/content-refresh` - Raw SKILL.md: https://api.skillmd.com/api/skills/teachskillofskills-ai/content-refresh/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Marketing & Growth - Author: teachskillofskills-ai (https://skillmd.com/u/teachskillofskills-ai) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/teachskillofskills-ai/content-refresh --- # Content Refresh Workflow Re-optimize existing content with updated research, current statistics, new sources, refreshed SEO keywords, and Phase 6.5 humanization — while preserving what's working and maintaining search rankings. ## When to Use Use `/contentforge:content-refresh` when: - Content is 6+ months old and needs updated stats/examples - Search rankings are declining (lost top 10 position) - Competitor content has surpassed yours - Product/service features have changed - Industry landscape has shifted - Content scored well originally (≥7.0) but needs freshening ## What This Command Does 1. **Load Existing Content** — Read the current .docx from wherever the brand's `tracking.backend` stores it (local filesystem, Airtable attachment, or Google Drive) 2. **Analyze What to Keep** — Identify evergreen sections, high-performing segments 3. **Research Updates** — Find current statistics, new sources, recent examples 4. **Selective Rewrite** — Update outdated sections, preserve working content 5. **Re-run Quality Gates** — Fact-check new claims, re-humanize, re-score 6. **SEO Preservation** — Maintain target keywords, internal links, meta structure 7. **Version Control** — Save as v1.1, v1.2 (never overwrite v1.0) ## Required Inputs **Existing Content:** - Google Drive URL or File ID - OR: Local .docx file path **Refresh Scope (select one):** - **Light Refresh** (20%): Update statistics, examples, citations only - **Medium Refresh** (50%): Rewrite intro/conclusion, update 3-5 sections, add new research - **Heavy Refresh** (80%): Complete rewrite using original as outline, keep only evergreen insights **Optional:** - New target keywords (if pivoting focus) - Sections to preserve (mark as "DO NOT EDIT") - Deadline (for priority ranking in batch) ## How to Use ### Basic Usage ``` /contentforge:content-refresh https://docs.google.com/document/d/XYZ123 ``` **Prompt:** "What refresh scope? (light / medium / heavy)" ### With Scope Specified ``` /contentforge:content-refresh https://docs.google.com/document/d/XYZ123 --scope=medium ``` ### Batch Refresh (multiple pieces) Run the refresh sheet through `/contentforge:batch-process` — point it at a sheet whose rows reference existing documents: ``` /contentforge:batch-process https://docs.google.com/spreadsheets/d/ABC123 ``` Sheet columns: `doc_url`, `refresh_scope`, `priority` ## What Happens ### Step 1: Content Analysis (2-3 minutes) - Load existing content from Google Drive - Extract metadata (original publish date, current word count, quality score) - Identify sections: intro, body paragraphs, conclusion, citations - **Evergreen Detection**: Flag sections that are timeless (definitions, principles, frameworks) - **Outdated Detection**: Flag statistics >12 months old, broken links, deprecated examples - Calculate "freshness score" (0-100, based on %outdated) **Output:** ``` Content Analysis Report ───────────────────────────────────────────────────── Title: "AI in Healthcare: 2025 Trends and Predictions" Original Publish: 2025-03-15 Current Word Count: 2,340 words Original Quality Score: 8.9/10 Freshness Score: 42/100 (Needs Refresh) Evergreen Sections (Keep): ✓ Para 2: Definition of AI in healthcare ✓ Para 5: Historical context (2010-2020) ✓ Para 8: Ethical considerations framework Outdated Sections (Update): ⚠ Para 1: Intro references "2025 predictions" (now outdated) ⚠ Para 3: Statistics from 2024 market report ⚠ Para 6: Example of startup acquired in 2025 ⚠ Para 10: Conclusion mentions "upcoming 2025 regulations" ⚠ Citations: 6/15 links are broken (404 errors) Recommendation: Medium Refresh (50% rewrite) ───────────────────────────────────────────────────── ``` ### Step 2: Research Phase (Targeted) - Run Phase 1 (Research Agent) focused ONLY on outdated sections - Search for: Current statistics (2026), new case studies, recent regulatory changes - Find replacement sources for broken citations - **Preserve existing sources** for evergreen content ### Step 3: Selective Rewrite - **Keep evergreen sections unchanged** (no rewrite) - **Update outdated sections** with new research - **Rewrite intro/conclusion** to reflect current year, updated predictions - **Maintain article structure** (same H2/H3 hierarchy) - **Preserve internal links** and brand-specific terminology ### Step 4: Re-run Quality Pipelines - **Phase 2 (Fact-Checker)**: Verify ONLY new claims and updated statistics - **Phase 4 (Validator)**: Check for hallucinations in rewritten sections - **Phase 5 (Structurer)**: Ensure refreshed content flows naturally with preserved sections - **Phase 6 (SEO)**: Preserve the original keyword **placements** (title, H1, first 100 words, 2-3 H2s, conclusion) and meta structure. Density is monitored for drift and reported — it is advisory, never a target to pad toward. - **Phase 6.5 (Humanizer)**: Re-humanize rewritten sections - **Phase 7 (Reviewer)**: Re-score (target: ±0.5 points from original score) ### Step 5: Version Control - Original: `Article-AI-Healthcare_v1.0.docx` (never modified) - Refresh: `Article-AI-Healthcare_v1.1.docx` (new version) - Track changes in metadata: "Refreshed 2026-02-17, updated 6 sections, added 4 new sources" ## Refresh Scopes ### Light Refresh (~20% rewrite, 8-12 min) **What Changes:** - Update statistics to current year - Replace 1-2 outdated examples - Fix broken citation links - Refresh intro sentence ("As of 2026..." instead of "In 2025...") - Re-run Phase 6.5 Humanizer only **What Stays:** - All structure (H2/H3 headings) - 80% of original paragraphs - All evergreen sections - Target keywords unchanged **Use Case:** Content is 6-12 months old, mostly accurate, just needs stats updated ### Medium Refresh (~50% rewrite, 15-20 min) **What Changes:** - Rewrite intro and conclusion completely - Update 40-60% of body paragraphs - Add 3-5 new sections for emerging trends - Replace 50% of citations with current sources - Re-run Phases 2, 4, 5, 6, 6.5, 7 **What Stays:** - Article structure (same H2 sections, order may change) - Evergreen definitions, frameworks, principles - Target keywords (may add 2-3 new secondary keywords) **Use Case:** Content is 12-24 months old, core thesis is valid but needs significant updates ### Heavy Refresh (~80% rewrite, 22-30 min) **What Changes:** - Complete rewrite using original as outline only - New research from scratch (Phase 1 full run) - Update target keywords based on current search intent - Add 5-10 new sections - Replace 80% of citations - Full 10-phase pipeline (same as new content) **What Stays:** - Core topic and brand voice - 1-2 evergreen sections (definitions, historical context) - SEO URL slug (to preserve backlinks) **Use Case:** Content is 24+ months old, industry has changed significantly, needs near-complete overhaul ## SEO Preservation Strategies ### Keyword Density Maintenance ``` Original keyword density: 2.3% for "AI in healthcare" Target for refresh: 2.0-2.6% (±0.3%) Action: Phase 6 monitors and adjusts rewritten sections ``` ### URL Slug Preservation ``` Original: /blog/ai-in-healthcare-2025-trends Refreshed: /blog/ai-in-healthcare-2025-trends (SAME URL) Title updates to: "AI in Healthcare: 2026 Trends and Predictions" ``` ### Internal Link Preservation - All internal links from original content are preserved - Add new internal links to related updated content - Never break existing internal link structure ### Meta Description Update ``` Original: "Explore AI in healthcare trends for 2025..." Refreshed: "Explore AI in healthcare trends for 2026..." (year updated) ``` ## Quality Scoring (Refresh vs. Original) **Target:** Refresh score should be within ±0.5 points of original **Example:** - Original: 8.9/10 (Content Quality: 9.2, Citations: 8.5, Brand: 9.0, SEO: 8.8, Readability: 9.0) - Refreshed: 9.1/10 (Content Quality: 9.3, Citations: 9.0, Brand: 9.0, SEO: 8.9, Readability: 9.2) - **Result: ✓ Within acceptable range** (+0.2 improvement) **If refresh scores <8.4** (<0.5 below original): - Flag for human review - Identify which dimension dropped (likely Citations or SEO) - Rerun Phase 2 (Fact-Checker) or Phase 6 (SEO Optimizer) ## Version Tracking ### Metadata in .docx ``` Document Properties: Title: AI in Healthcare: 2026 Trends and Predictions Version: 1.1 Original Publish Date: 2025-03-15 Last Refresh: 2026-02-17 Refresh Scope: Medium (50%) Sections Updated: 6/12 New Sources Added: 4 Quality Score: 9.1/10 (was 8.9/10) Refreshed By: ContentForge ``` ### Filename Convention ``` Original: Article-AI-Healthcare_v1.0.docx 1st Refresh: Article-AI-Healthcare_v1.1.docx 2nd Refresh: Article-AI-Healthcare_v1.2.docx Major Rewrite: Article-AI-Healthcare_v2.0.docx (Heavy Refresh) ``` ### Google Drive Organization ``` ContentForge Output/ └── Article-AI-Healthcare/ ├── Article-AI-Healthcare_v1.0.docx (Original, 2025-03-15) ├── Article-AI-Healthcare_v1.1.docx (Refresh, 2026-02-17) └── refresh-comparison-report.txt ``` ## Comparison Report After refresh, generate side-by-side comparison: ``` ═══════════════════════════════════════════════════════════════ Content Refresh Comparison Report ═══════════════════════════════════════════════════════════════ Article: AI in Healthcare: 2026 Trends and Predictions Refresh Date: 2026-02-17 Scope: Medium (50%) Metrics Comparison: ───────────────────────────────────────────────────────────── Metric │ Original (v1.0) │ Refreshed (v1.1) │ Change ───────────────────────────────────────────────────────────── Word Count │ 2,340 │ 2,485 │ +145 (+6%) Quality Score │ 8.9/10 │ 9.1/10 │ +0.2 Citations │ 15 │ 19 │ +4 Broken Links │ 6 (40%) │ 0 (0%) │ -6 (fixed) Keyword Density │ 2.3% │ 2.4% │ +0.1% Readability (Grade) │ 11.2 │ 10.8 │ -0.4 (easier) Freshness Score │ 42/100 │ 95/100 │ +53 Sections Changed: ───────────────────────────────────────────────────────────── ✓ Introduction (completely rewritten) ✓ Section 2: Statistics updated (2024 → 2026 data) ✓ Section 4: New case study added (2026 example) ✓ Section 7: Regulatory update (new FDA guidelines) ✓ Section 9: Predictions updated (2026-2028 outlook) ✓ Conclusion (completely rewritten) Sections Preserved (Evergreen): ───────────────────────────────────────────────────────────── Section 1: AI Healthcare Definition Section 3: Historical Context (2010-2020) Section 6: Ethical Framework SEO Impact Assessment: ───────────────────────────────────────────────────────────── Target Keyword: "AI in healthcare" Density: 2.3% → 2.4% ✓ (within target) First mention: Para 1, Sentence 2 ✓ (unchanged) Meta Title: "AI in Healthcare: 2026 Trends..." Length: 42 chars ✓ (optimal) Updated year: 2025 → 2026 ✓ URL Slug: /blog/ai-in-healthcare-2025-trends Preserved: ✓ (maintains backlinks) Internal Links: 8 preserved, 3 added ✓ Estimated SEO Impact: +5-10% traffic (fresher content, fixed broken links) ═══════════════════════════════════════════════════════════════ Recommendation: Publish refreshed version, monitor rankings for 2 weeks ═══════════════════════════════════════════════════════════════ ``` ## Batch Content Refresh ### Use Case: Quarterly Content Audit Agency has 50 blog posts, wants to refresh top 20 performers that are 12+ months old. **Step 1: Prepare Refresh Sheet** ```csv doc_url,refresh_scope,priority,notes https://docs.google.com/.../article-1,medium,1,Rankings dropped from #3 to #7 https://docs.google.com/.../article-2,light,2,Just needs stat updates https://docs.google.com/.../article-3,heavy,3,Topic outdated, needs rewrite ... ``` **Step 2: Run Batch Refresh** ``` /contentforge:batch-process https://docs.google.com/spreadsheets/d/ABC123 ``` **Step 3: Process (standard batch orchestration)** - **Sequential — one refresh pipeline at a time.** There is no concurrency. - Queue order is set by the `priority` column - Every phase of every piece is checkpointed, so an interrupted batch resumes where it stopped rather than restarting - Completion report with before/after scores Total time scales roughly with the number of pieces and their refresh scope; batching buys queue management and resumability, not a speedup. ## Integration with Other Skills **Before Refresh:** - `/contentforge:cf-audit` — Identify which content needs refreshing (freshness scores, declining candidates). **v4.0 file contract:** when the user says "refresh what the audit found" (or `--from-audit`), read the recorded candidates rather than remembering them: ```bash python ${CLAUDE_PLUGIN_ROOT}/scripts/audit-ledger.py latest --brand ``` Take `pieces` ranked by `refresh_priority`; each row's `recommended_scope` (light/medium/heavy/retire) is this skill's scope input. Exit 1 means no recorded audit exists — say so and point at `/contentforge:cf-audit`; never invent candidates. - `/contentforge:cf-brief` — Re-run keyword + competitor research on the topic to see what the SERP now rewards and what competitors have added since you published **After Refresh:** - `/contentforge:publish` — Push updated content to WordPress/Webflow - `/contentforge:cf-analytics` — Track refresh quality scores over time ## Limitations - Content not originally created by ContentForge has **no baseline quality score**, so the refresh is graded absolutely rather than as a delta — the "±0.5 of original" criterion below does not apply to imported pieces handed over from `/contentforge:cf-audit` - Heavy Refresh (80%) is almost same time as new content (use sparingly) - Requires the original .docx to be retrievable from the brand's configured tracking backend (can't refresh from published URLs alone) ## Success Criteria **Good Refresh:** - Quality score within ±0.5 of original - Freshness score improves to 85-100 - All broken links fixed - Keyword placements preserved (density drift reported as advisory) - SEO rankings stable or improve within 2-4 weeks **Bad Refresh (requires redo):** - Quality score drops >1.0 point - A keyword placement slot (title, H1, first 100 words, H2s, conclusion) lost its primary keyword - Internal links broken - Brand voice inconsistency ## Agents Used The refresh reuses the canonical 10-phase pipeline agents (Reviewer is Phase 7, Output Manager is Phase 8): - **Heavy Refresh** — full pipeline, same as new content (Step 0.5 + Phases 1-8) - **Medium Refresh** — Phases 2 (Fact-Checker), 4 (Scientific Validator), 5 (Structurer), 6 (SEO/AEO/GEO), 6.5 (Humanizer), 7 (Reviewer), 8 (Output Manager) - **Light Refresh** — Phases 6.5 (Humanizer), 7 (Reviewer), 8 (Output Manager) ## Related Skills - `/contentforge:batch-process` — Create or refresh content as a sequential, checkpointed queue - `/contentforge:cf-variants` — A/B test refreshed vs. original elements - `/contentforge:cf-audit` — Find refresh candidates across the whole library --- **Value:** Preserves SEO equity (URL, internal links, keyword placements) while extending content lifespan