Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when the user asks to "find keywords", "挖词", or "搜什么词"; prioritizes search volume, keyword difficulty, intent, and topic clusters from provided or connected data. Not for competitor-relative coverage gaps — use content-gap-analysis. 关键词研究/内容选题
.claude/skills/itamarzand88-keyword-research/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | 67% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 36% | 0% |
<!-- source: seo-keyword-research — https://raw.githubusercontent.com/aaron-he-zhu/seo-geo-claude-skills/main/research/keyword-research/SKILL.md -->
Discovers, scores, and clusters keywords for SEO and GEO planning.
Research keywords for [topic/product/service]What keywords is [competitor URL] ranking for that I should target?Expected output: a prioritized keyword brief plus the standard handoff summary for memory/research/.
memory/hot-cache.md, memory/open-loops.md, and memory/research/.> Emit the standard shape from skill-contract.md §Handoff Summary Format.
Optional integrations: ~~SEO tool, ~~search console. Without tools, ask for seed keywords, audience, goals, and any known metrics. See CONNECTORS.md.
Zero-dependency local helper (no tool needed): python3 scripts/connectors/suggest.py "<seed>" --expand harvests free keyword ideas from Google Autocomplete (⚠️ unofficial endpoint). Search volume / difficulty still needs ~~SEO tool or own Search Console data. See scripts/connectors/README.md.
When a user requests keyword research, run eight phases and announce each as [Phase X/8: Name]:
Opportunity = (Volume × Intent Value) / Difficulty, with Intent Value 1 / 1 / 2 / 3.Label every metric Measured (tool/export), User-provided, or Estimated (model inference); never present an estimate as measured; if a required metric is unavailable, mark it N/A — do not invent it.
Quality bar: every recommendation includes at least one specific number. Rewrite generic advice into a concrete keyword + volume + difficulty + reason.
> Reference: See references/instructions-detail.md for the full 8-phase templates, expansion patterns, intent table, difficulty tiers, opportunity matrix, GEO indicators, cluster template, actionable-vs-generic examples, and advanced usage.
Example outcome: 150+ keywords analyzed, 23 high-priority opportunities, ~45K/month traffic potential across 3 focus areas. See the full sample in references/example-report.md.
Intent mapping, seasonal analysis, competitor gaps, and local keyword workflows live in references/instructions-detail.md.
Start with seeds, respect intent, cluster tightly, prioritize quick wins, and review quarterly. Full notes live in references/instructions-detail.md.
Write path: memory/research/keyword-research/YYYY-MM-DD-<topic>.md; promote durable keyword priorities to memory/hot-cache.md. See Skill Contract §Save Results Template.
Primary: competitor-analysis. Also: content-gap-analysis and serp-analysis.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | fail→pass | 11,465 | 9,659 | -16% | 1 | 1 | 0% | 2,287 | 3,810 | +67% | 0 | 0 | — |
case-01 | fail→pass | 30,795 | 27,304 | -11% | 1 | 1 | 0% | 5,930 | 7,192 | +21% | 0 | 0 | — |
case-02 | fail→pass | 79,858 | 18,920 | -76% | 1 | 1 | 0% | 4,051 | 5,714 | +41% | 0 | 0 | — |
case-03 | fail→pass | 28,452 | 26,304 | -8% | 1 | 1 | 0% | 6,211 | 7,935 | +28% | 0 | 0 | — |
case-04 | pass→pass | 19,682 | 18,583 | -6% | 1 | 1 | 0% | 3,626 | 5,290 | +46% | 0 | 0 | — |
case-05 | pass→pass | 6,061 | 6,965 | +15% | 1 | 1 | 0% | 1,284 | 3,005 | +134% | 0 | 0 | — |
case-06 | fail→fail | 12,839 | 24,621 | +92% | 1 | 1 | 0% | 2,350 | 6,591 | +180% | 0 | 0 | — |
case-07 | fail→pass | 10,336 | 6,066 | -41% | 1 | 1 | 0% | 2,250 | 3,064 | +36% | 0 | 0 | — |
case-08 | pass→pass | 10,235 | 4,096 | -60% | 1 | 1 | 0% | 2,044 | 2,498 | +22% | 0 | 0 | — |
case-10 | pass→pass | 7,082 | 7,259 | +2% | 1 | 1 | 0% | 1,655 | 3,287 | +99% | 0 | 0 | — |
case-11 | fail→pass | 10,047 | 6,740 | -33% | 1 | 1 | 0% | 2,087 | 2,980 | +43% | 0 | 0 | — |
case-12 | fail→pass | 6,978 | 5,391 | -23% | 1 | 1 | 0% | 1,291 | 2,834 | +120% | 0 | 0 | — |
case-13 | pass→pass | 6,217 | 3,123 | -50% | 1 | 1 | 0% | 1,288 | 2,325 | +81% | 0 | 0 | — |
case-14 | pass→pass | 8,894 | 2,683 | -70% | 1 | 1 | 0% | 1,581 | 2,188 | +38% | 0 | 0 | — |
case-15 | fail→pass | 13,068 | 4,121 | -68% | 1 | 1 | 0% | 3,314 | 2,487 | -25% | 0 | 0 | — |
case-16 | fail→pass | 6,002 | 2,485 | -59% | 1 | 1 | 0% | 1,281 | 2,292 | +79% | 0 | 0 | — |
case-17 | fail→pass | 8,195 | 1,908 | -77% | 1 | 1 | 0% | 1,645 | 2,133 | +30% | 0 | 0 | — |
case-18 | fail→pass | 9,225 | 3,204 | -65% | 1 | 1 | 0% | 1,692 | 2,392 | +41% | 0 | 0 | — |
case-19 | pass→fail | 3,727 | 1,232 | -67% | 1 | 1 | 0% | 742 | 1,975 | +166% | 0 | 0 | — |
case-20 | fail→pass | 7,323 | 4,315 | -41% | 1 | 1 | 0% | 1,406 | 2,643 | +88% | 0 | 0 | — |
case-21 | pass→pass | 6,563 | 1,486 | -77% | 1 | 1 | 0% | 1,166 | 1,997 | +71% | 0 | 0 | — |
case-22 | fail→pass | 9,248 | 3,272 | -65% | 1 | 1 | 0% | 1,803 | 2,332 | +29% | 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 +55 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
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.