Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Triage technical SEO findings from Ahrefs Site Audit (and similar crawlers) by SEO impact, not just severity. Use this skill when reviewing crawl results, prioritizing technical fixes, scoping a technical SEO sprint, or after running any site-wide crawl. Triggers on site audit results, technical fix list, crawl errors, technical SEO triage, prioritize technical issues, what should we fix first, broken links, redirect chains. Also triggers when a long list of crawler issues is creating decision p
.claude/skills/rampstackco-seo-site-health-audit/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 137% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 117% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 88% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 94% | 0% |
Triage technical SEO findings from Ahrefs Site Audit (or any equivalent crawler) by impact on rankings and traffic, not by raw issue count or severity label. Stack-agnostic. Produces a prioritized backlog of fixes mapped to business impact.
seo-traffic-diagnosis)seo-onpage)seo-technical)Crawlers label issues "critical", "warning", "notice". These labels are useful but not sufficient. Two issues both labeled "critical" can have wildly different actual impact on the property.
Triage on three axes.
Does the issue affect a page that drives meaningful organic traffic, or one that does not?
A "critical" issue on a tag archive with zero traffic is lower priority than a "warning" on a top-10-traffic landing page.
Tier the affected URLs:
A fix on Tier 1 is worth 10x a fix on Tier 3 in most cases.
Does the issue actually move rankings or traffic? Some "critical" issues are aesthetic or theoretical.
Grade every issue HIGH, MEDIUM, or LOW. references/issue-impact-table.md owns the grade for each issue type and the definition of each grade. Look the issue up there instead of judging it fresh, and use the grade it gives.
The issue families the reference covers, in lookup order: indexability, crawlability, renderability, Core Web Vitals, structured data, internal link integrity, hreflang, on-page elements, images, security and trust. It also carries a park-or-skip list for the findings that rarely deserve dedicated work.
Grades vary inside a family. Indexability rows are mostly HIGH; internal link integrity runs MEDIUM down to LOW row by row. Grade by row, never by family.
Fix HIGH-grade issues on Tier 1 URLs first.
Some issues are 5-minute fixes. Some are multi-sprint projects.
Combine the three axes into a priority score.
Mechanism here is the HIGH / MEDIUM / LOW grade the issue-impact table gives the issue. All three grades have rows: the reference grades a large share of its issue types MEDIUM, and a MEDIUM issue on a Tier 1 page still needs a band.
| Tier | Mechanism | Effort | Priority | | --- | --- | --- | --- | | Tier 1 | HIGH | S | P0 (do this week) | | Tier 1 | HIGH | M | P1 (do this sprint) | | Tier 1 | HIGH | L | P2 (plan as project) | | Tier 1 | MEDIUM | S | P1 | | Tier 1 | MEDIUM | M | P2 | | Tier 1 | MEDIUM | L | P3 | | Tier 1 | LOW | S | P3 (batch when convenient) | | Tier 1 | LOW | M-L | Park unless evidence emerges | | Tier 2 | HIGH | S | P1 | | Tier 2 | HIGH | M-L | P2 | | Tier 2 | MEDIUM | S | P2 | | Tier 2 | MEDIUM | M-L | P3 | | Tier 3 | HIGH | S | P3 | | Tier 3 | Anything else | Anything | Park |
The MEDIUM rows follow one rule: a MEDIUM issue sits one band below the HIGH issue at the same tier and effort. Rows the reference grades LOW-MEDIUM take the MEDIUM row when the URL is Tier 1 and the LOW row otherwise.
P0-P1 work earns the team's attention. P2 goes on the roadmap. P3 batches into routine maintenance. Park is honest about deprioritization.
references/issue-impact-table.md.seo-audit-orchestration.These patterns commonly produce outsized impact when fixed:
When you spot one of these, prioritize even if individual issues look small. The cluster impact is large.
A site health triage document with:
Length: 5-12 pages plus a backlog spreadsheet.
This skill's output depends on data, measurements, or tool results it cannot generate on its own. When a required input, tool, or data source is unavailable or unverifiable, the sanctioned output is the deliverable with the gap stated: what was needed, what was actually obtained or verified, and which parts of the output are affected. Fabricating, estimating, or interpolating a required number to complete the deliverable is never sanctioned. A stated gap is a complete answer.
references/issue-impact-table.md - Mapping table of common crawler issues to mechanism grade and typical fix effort, plus cluster signals and the park-or-skip list. This file is the source of truth for every issue grade; the triage matrix above turns a grade plus a tier plus an effort into a priority band.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 36,027 | 35,847 | -0% | 1 | 1 | 0% | 6,291 | 8,556 | +36% | 0 | 0 | — |
case-02 | fail→fail | 35,943 | 34,423 | -4% | 1 | 1 | 0% | 6,291 | 8,556 | +36% | 0 | 0 | — |
case-03 | fail→fail | 34,821 | 35,155 | +1% | 1 | 1 | 0% | 6,276 | 8,541 | +36% | 0 | 0 | — |
case-08 | pass→pass | 11,151 | 3,834 | -66% | 1 | 1 | 0% | 1,781 | 2,902 | +63% | 0 | 0 | — |
case-04 | fail→pass | 9,318 | 8,448 | -9% | 1 | 1 | 0% | 1,577 | 3,745 | +137% | 0 | 0 | — |
case-05 | pass→pass | 14,204 | 13,140 | -7% | 1 | 1 | 0% | 2,425 | 4,457 | +84% | 0 | 0 | — |
case-06 | pass→pass | 15,041 | 14,694 | -2% | 1 | 1 | 0% | 2,677 | 4,543 | +70% | 0 | 0 | — |
case-07 | pass→pass | 11,274 | 7,249 | -36% | 1 | 1 | 0% | 1,886 | 3,467 | +84% | 0 | 0 | — |
case-09 | pass→pass | 15,688 | 7,195 | -54% | 1 | 1 | 0% | 2,490 | 3,487 | +40% | 0 | 0 | — |
case-10 | pass→pass | 6,041 | 2,949 | -51% | 1 | 1 | 0% | 980 | 2,772 | +183% | 0 | 0 | — |
case-11 | fail→pass | 9,090 | 5,842 | -36% | 1 | 1 | 0% | 1,487 | 3,225 | +117% | 0 | 0 | — |
case-12 | fail→pass | 11,031 | 5,362 | -51% | 1 | 1 | 0% | 1,668 | 3,137 | +88% | 0 | 0 | — |
case-13 | pass→pass | 14,819 | 12,526 | -15% | 1 | 1 | 0% | 2,277 | 4,274 | +88% | 0 | 0 | — |
case-14 | pass→pass | 12,225 | 11,159 | -9% | 1 | 1 | 0% | 2,122 | 4,108 | +94% | 0 | 0 | — |
case-15 | pass→pass | 13,024 | 10,205 | -22% | 1 | 1 | 0% | 2,164 | 3,958 | +83% | 0 | 0 | — |
case-16 | pass→pass | 23,839 | 12,660 | -47% | 1 | 1 | 0% | 4,102 | 4,642 | +13% | 0 | 0 | — |
case-17 | pass→pass | 11,021 | 9,827 | -11% | 1 | 1 | 0% | 1,747 | 3,817 | +118% | 0 | 0 | — |
case-18 | pass→pass | 15,187 | 10,911 | -28% | 1 | 1 | 0% | 2,474 | 3,876 | +57% | 0 | 0 | — |
case-19 | pass→pass | 15,082 | 14,894 | -1% | 1 | 1 | 0% | 2,222 | 4,543 | +104% | 0 | 0 | — |
case-20 | pass→pass | 9,916 | 4,673 | -53% | 1 | 1 | 0% | 1,450 | 3,104 | +114% | 0 | 0 | — |
case-21 | fail→pass | 12,211 | 6,529 | -47% | 1 | 1 | 0% | 1,696 | 3,288 | +94% | 0 | 0 | — |
case-22 | pass→pass | 12,247 | 16,326 | +33% | 1 | 1 | 0% | 1,805 | 4,609 | +155% | 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 +23 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.