Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when planning to rank for a specific keyword. The agent Googles it, reads the top 10, classifies intent, reads the top 3 competitor pages, and produces a 90-day ranking plan with intent, SERP analysis, and content recommendations.
.claude/skills/inhouseseo-keyword-deep-dive/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 72% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -4% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 178% | 0% |
A complete picture of a single keyword's opportunity — intent, competition, what it takes to rank, and a specific 90-day plan. The agent does all the research itself. No keyword tool required.
Target keyword (required). Optionally: a URL if you already have a page targeting this keyword.
If the user didn't provide a keyword, ask for it before proceeding.
You are a senior SEO strategist specializing in keyword intelligence and SERP analysis.
Google the target keyword. Read the top 10 results in detail, the top 3 in full.
Note for each top 3:
Classify dominant intent: Informational / Commercial Investigation / Transactional / Navigational.
Zero-click risk assessment:
What SERP features are active?
Can you tell if the SERP is stable or turbulent?
Fetch and read the top 3 pages. For each:
For each of the top 3 competitors:
Specific subtopics or angles that top-ranking pages don't cover well. These are where a new entrant can differentiate.
If the user has no existing page for this keyword:
Quick assessment:
Content requirements:
If the user already has a page ranking:
Position diagnosis:
Optimization plan:
Propose 2 title tag options (max 60 chars) and 1 meta description (max 160 chars). Each with reasoning.
Need a full content brief to execute the ranking plan? Use content-brief with this keyword as context.
Load from references/ only when the step calls for them.
ctr-benchmarks-by-position.md — the full 2026 position → CTR curve with AI-Overview-adjusted numbers (Step 2, when estimating traffic potential)zero-click-and-aio-impact.md — how AI Overviews, featured snippets, and zero-click SERPs change the ranking math (Step 2, for the zero-click risk line)serp-features-recognition.md — checklist for detecting every SERP feature visually (Step 3, if uncertain which features are active)serp-volatility-heuristics.md — how to read SERP volatility without access to a rank tracker (Step 4, when volatility is ambiguous)difficulty-from-serp-signals.md — how to estimate KD from reading the top 10 instead of a tool score (Step 5, when classifying Easy/Moderate/Hard)intent-matching.md — the full Informational / Commercial / Transactional / Navigational decision framework (Step 2, when intent is mixed)| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→pass | 21,193 | 24,215 | +14% | 1 | 1 | 0% | 3,117 | 5,371 | +72% | 0 | 0 | — |
case-01 | fail→fail | 23,810 | 28,804 | +21% | 1 | 1 | 0% | 3,710 | 6,077 | +64% | 0 | 0 | — |
case-03 | fail→pass | 25,879 | 33,073 | +28% | 1 | 1 | 0% | 3,892 | 6,041 | +55% | 0 | 0 | — |
case-04 | fail→pass | 17,338 | 5,615 | -68% | 1 | 1 | 0% | 2,624 | 2,530 | -4% | 0 | 0 | — |
case-05 | fail→pass | 19,506 | 5,579 | -71% | 1 | 1 | 0% | 3,122 | 2,419 | -23% | 0 | 0 | — |
case-06 | fail→fail | 16,826 | 6,736 | -60% | 1 | 1 | 0% | 2,450 | 2,628 | +7% | 0 | 0 | — |
case-07 | pass→pass | 14,353 | 8,763 | -39% | 1 | 1 | 0% | 2,050 | 3,012 | +47% | 0 | 0 | — |
case-08 | pass→pass | 5,644 | 4,646 | -18% | 1 | 1 | 0% | 799 | 2,352 | +194% | 0 | 0 | — |
case-09 | pass→pass | 14,301 | 13,407 | -6% | 1 | 1 | 0% | 2,200 | 3,813 | +73% | 0 | 0 | — |
case-10 | pass→pass | 18,864 | 26,763 | +42% | 1 | 1 | 0% | 2,632 | 5,566 | +111% | 0 | 0 | — |
case-11 | pass→pass | 16,588 | 21,587 | +30% | 1 | 1 | 0% | 2,306 | 4,805 | +108% | 0 | 0 | — |
case-12 | pass→pass | 14,278 | 18,440 | +29% | 1 | 1 | 0% | 2,181 | 4,353 | +100% | 0 | 0 | — |
case-13 | fail→fail | 8,711 | 21,307 | +145% | 1 | 1 | 0% | 1,626 | 4,551 | +180% | 0 | 0 | — |
case-14 | fail→pass | 11,087 | 18,862 | +70% | 1 | 1 | 0% | 1,697 | 4,719 | +178% | 0 | 0 | — |
case-15 | pass→pass | 11,191 | 16,875 | +51% | 1 | 1 | 0% | 1,701 | 4,224 | +148% | 0 | 0 | — |
case-16 | pass→pass | 10,578 | 21,177 | +100% | 1 | 1 | 0% | 1,571 | 4,972 | +216% | 0 | 0 | — |
case-17 | fail→fail | 14,415 | 21,817 | +51% | 1 | 1 | 0% | 2,162 | 4,680 | +116% | 0 | 0 | — |
case-18 | fail→fail | 17,535 | 11,705 | -33% | 1 | 1 | 0% | 2,647 | 3,505 | +32% | 0 | 0 | — |
case-19 | fail→pass | 15,830 | 19,095 | +21% | 1 | 1 | 0% | 2,569 | 4,674 | +82% | 0 | 0 | — |
case-20 | fail→pass | 12,850 | 21,680 | +69% | 1 | 1 | 0% | 1,984 | 4,349 | +119% | 0 | 0 | — |
case-21 | fail→pass | 14,704 | 5,138 | -65% | 1 | 1 | 0% | 2,310 | 2,359 | +2% | 0 | 0 | — |
case-22 | fail→pass | 8,834 | 3,084 | -65% | 1 | 1 | 0% | 1,372 | 2,086 | +52% | 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 +41 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.