Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when the user asks to "write SEO content"; drafts new posts, articles, and landing pages with keywords, headers, snippets, and evidence boundaries. Not for AI-citation/GEO readiness scoring — use geo-content-optimizer; not for updating decaying existing content — use content-refresher. SEO文章写作/内容优化
.claude/skills/itamarzand88-seo-content-writer/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 6 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 69% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 50% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 83% | 0% |
<!-- source: seo-content-writer — https://raw.githubusercontent.com/aaron-he-zhu/seo-geo-claude-skills/main/build/seo-content-writer/SKILL.md -->
Creates SEO content that aligns with search intent, integrates keywords naturally, and stays usable for readers.
Write an SEO-optimized article about [topic] targeting the keyword [keyword]Here's my content brief: [brief]. Write SEO-optimized content following this outline.Expected output: a ready-to-use draft plus the standard handoff summary for memory/content/.
memory/hot-cache.md and memory/open-loops.md; propose durable decisions as pending-decision items.> Emit the standard shape from skill-contract.md §Handoff Summary Format.
Use ~~SEO tool and ~~search console when connected; otherwise ask for keywords, intent, and competitors. See CONNECTORS.md.
When a user requests SEO content, run these nine steps:
Any factual claim, statistic, or quote that requires a source must be cited or explicitly flagged [needs source]; never invent figures, studies, dates, or attributions to fill a gap.
Quality bar: before handing off, confirm the draft passes — (1) intent match: a reader with the target query gets their answer above the fold; (2) keyword placement reads naturally (no stuffing) in title, H1, first 100 words, and one H2; (3) structure is scannable (H2/H3, lists, one snippet-ready block); (4) zero fabricated facts — every source-needing claim is cited or [needs source]. If any item fails, fix it or report it in the handoff, do not ship silently.
> Reference: See Instructions Detail for the compact workflow, pre-write checklist, issue-classification rules, and self-check format.
Sample outcome: a keyword-led H1, optimized meta description, clear H2 structure, FAQ section, and a brief Changes Made block after the self-check. See references/seo-writing-checklist.md for the copy-start checklist and article template.
Quick-start patterns for how-to guides, comparisons, listicles, pillar pages, reviews, and FAQ pages live in references/content-structure-templates.md.
Match intent, front-load value, support claims with evidence, and write for humans before optimizing for the SERP.
On user confirmation, save to memory/content/YYYY-MM-DD-<topic>.md — see Skill Contract §Save Results Template.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→pass | 20,108 | 18,726 | -7% | 1 | 1 | 0% | 3,161 | 4,520 | +43% | 0 | 0 | — |
case-03 | fail→pass | 27,486 | 22,496 | -18% | 1 | 1 | 0% | 4,870 | 5,837 | +20% | 0 | 0 | — |
case-17 | pass→pass | 16,354 | 19,773 | +21% | 1 | 1 | 0% | 2,509 | 4,769 | +90% | 0 | 0 | — |
case-04 | fail→fail | 20,358 | 22,210 | +9% | 1 | 1 | 0% | 3,640 | 5,414 | +49% | 0 | 0 | — |
case-05 | pass→pass | 7,261 | 5,396 | -26% | 1 | 1 | 0% | 1,155 | 2,423 | +110% | 0 | 0 | — |
case-01 | fail→fail | 25,306 | 23,969 | -5% | 1 | 1 | 0% | 4,470 | 6,463 | +45% | 0 | 0 | — |
case-06 | fail→fail | 18,755 | 19,927 | +6% | 1 | 1 | 0% | 3,004 | 4,973 | +66% | 0 | 0 | — |
case-07 | pass→pass | 18,523 | 28,563 | +54% | 1 | 1 | 0% | 3,121 | 5,106 | +64% | 0 | 0 | — |
case-08 | fail→fail | 16,917 | 18,223 | +8% | 1 | 1 | 0% | 2,747 | 4,694 | +71% | 0 | 0 | — |
case-09 | fail→pass | 18,686 | 21,517 | +15% | 1 | 1 | 0% | 2,991 | 5,066 | +69% | 0 | 0 | — |
case-10 | fail→pass | 18,262 | 17,396 | -5% | 1 | 1 | 0% | 2,940 | 4,422 | +50% | 0 | 0 | — |
case-11 | pass→pass | 16,447 | 19,278 | +17% | 1 | 1 | 0% | 2,469 | 4,402 | +78% | 0 | 0 | — |
case-12 | pass→pass | 15,559 | 22,994 | +48% | 1 | 1 | 0% | 2,470 | 5,146 | +108% | 0 | 0 | — |
case-13 | fail→fail | 20,455 | 22,457 | +10% | 1 | 1 | 0% | 2,945 | 4,999 | +70% | 0 | 0 | — |
case-14 | pass→pass | 33,583 | 22,661 | -33% | 1 | 1 | 0% | 3,774 | 5,271 | +40% | 0 | 0 | — |
case-15 | pass→pass | 17,536 | 17,315 | -1% | 1 | 1 | 0% | 3,158 | 4,436 | +40% | 0 | 0 | — |
case-16 | fail→pass | 14,835 | 17,324 | +17% | 1 | 1 | 0% | 2,387 | 4,363 | +83% | 0 | 0 | — |
case-18 | fail→fail | 14,215 | 19,823 | +39% | 1 | 1 | 0% | 2,549 | 4,772 | +87% | 0 | 0 | — |
case-19 | pass→pass | 12,964 | 11,264 | -13% | 1 | 1 | 0% | 2,306 | 3,511 | +52% | 0 | 0 | — |
case-20 | fail→pass | 13,482 | 16,497 | +22% | 1 | 1 | 0% | 2,435 | 4,350 | +79% | 0 | 0 | — |
case-21 | pass→pass | 21,267 | 22,870 | +8% | 1 | 1 | 0% | 2,894 | 5,574 | +93% | 0 | 0 | — |
case-22 | pass→pass | 29,823 | 28,532 | -4% | 1 | 1 | 0% | 5,352 | 6,606 | +23% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +27 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.