Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →SEO strategy for technical queries and developer audiences. Covers keyword research for "how to X in language" queries, error message SEO, Stack Overflow-style content, technical long-tail keywords, and competing with official documentation sites. Use when asked about: - SEO for...
.claude/skills/sickn33-developer-seo/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | 59% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 83% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-23 | ✓→✗ | ▼ Worse | 58% | 0% |
| case-19 | ✓→✓ | ＝ Same ✓ | 105% | 0% |
Use this skill when you need sEO strategy for technical queries and developer audiences. Covers keyword research for "how to X in language" queries, error message SEO, Stack Overflow-style content, technical long-tail keywords, and competing with official documentation sites. Use when asked about: - SEO for...
Developer SEO differs fundamentally from traditional SEO. Developers search with precise technical intent—error messages, API questions, "how to X in Y language" queries. They bounce immediately from thin content and respect sites that actually solve problems. Your competition isn't other marketing sites; it's Stack Overflow, official docs, and GitHub issues.
This skill covers SEO strategies that work for technical audiences without compromising on substance.
Developers search differently than general audiences:
Query patterns:
Behavioral signals:
Technical long-tail keywords have lower volume but extremely high intent. A developer searching "axios interceptor refresh token react" knows exactly what they need.
Research approaches:
Error messages are SEO gold—developers copy-paste them directly into search.
Strategy:
Content structure for error pages:
Title: [Exact Error Message] - How to Fix
## The Error
[Full error message and where it appears]
## Quick Fix
[The solution that works in most cases]
## Why This Happens
[Brief technical explanation]
## Other Solutions
[Alternative fixes for edge cases]
## Related Errors
[Links to similar issues]Official docs have domain authority advantages but often have weaknesses:
Where docs often fail:
Your opportunities:
Structure for technical how-to content:
markdown# How to [Action] in [Technology] ## Prerequisites - What you need before starting - Required versions/dependencies ## Quick Version (TL;DR) - Code snippet that works for common case ## Step-by-Step 1. Step with explanation 2. Step with code example 3. Step with expected output ## Complete Example [Full working code] ## Common Issues - Problem 1: Solution - Problem 2: Solution ## Next Steps [What to learn next]
Developers actively search "Tool A] vs Tool B]" when evaluating options.
Guidelines:
In-depth tutorials build topical authority and capture multiple related queries.
Planning approach:
Google can read and understand code. Optimize for it:
<code>, <pre>)Developers expect fast sites. They also often use ad blockers and privacy tools.
Priorities:
Good IA helps both users and search engines:
High-quality technical backlinks matter more than quantity.
Sources that work:
What doesn't work:
Developer content becomes outdated quickly:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-19 | pass→pass | 17,014 | 20,678 | +22% | 1 | 1 | 0% | 2,383 | 4,888 | +105% | 0 | 0 | — |
case-01 | pass→pass | 15,669 | 10,065 | -36% | 1 | 1 | 0% | 2,701 | 3,724 | +38% | 0 | 0 | — |
case-02 | pass→pass | 17,800 | 9,640 | -46% | 1 | 1 | 0% | 2,902 | 3,516 | +21% | 0 | 0 | — |
case-03 | fail→fail | 13,230 | 13,527 | +2% | 1 | 1 | 0% | 2,192 | 4,000 | +82% | 0 | 0 | — |
case-04 | pass→pass | 15,953 | 16,780 | +5% | 1 | 1 | 0% | 2,616 | 4,409 | +69% | 0 | 0 | — |
case-05 | pass→pass | 15,473 | 14,708 | -5% | 1 | 1 | 0% | 2,498 | 4,434 | +78% | 0 | 0 | — |
case-06 | pass→pass | 13,120 | 13,963 | +6% | 1 | 1 | 0% | 2,188 | 4,131 | +89% | 0 | 0 | — |
case-07 | pass→pass | 14,974 | 12,527 | -16% | 1 | 1 | 0% | 2,337 | 3,876 | +66% | 0 | 0 | — |
case-08 | fail→pass | 16,131 | 14,417 | -11% | 1 | 1 | 0% | 2,651 | 4,227 | +59% | 0 | 0 | — |
case-09 | fail→fail | 12,887 | 11,861 | -8% | 1 | 1 | 0% | 2,055 | 3,864 | +88% | 0 | 0 | — |
case-10 | pass→pass | 15,728 | 11,896 | -24% | 1 | 1 | 0% | 2,635 | 3,869 | +47% | 0 | 0 | — |
case-11 | fail→fail | 17,014 | 13,945 | -18% | 1 | 1 | 0% | 2,588 | 4,311 | +67% | 0 | 0 | — |
case-12 | pass→pass | 17,524 | 16,579 | -5% | 1 | 1 | 0% | 2,645 | 4,572 | +73% | 0 | 0 | — |
case-13 | pass→pass | 16,799 | 16,204 | -4% | 1 | 1 | 0% | 2,467 | 4,563 | +85% | 0 | 0 | — |
case-14 | pass→pass | 15,759 | 15,131 | -4% | 1 | 1 | 0% | 2,531 | 4,331 | +71% | 0 | 0 | — |
case-15 | fail→pass | 10,851 | 7,704 | -29% | 1 | 1 | 0% | 1,677 | 3,070 | +83% | 0 | 0 | — |
case-16 | pass→pass | 11,984 | 7,508 | -37% | 1 | 1 | 0% | 1,893 | 3,040 | +61% | 0 | 0 | — |
case-17 | pass→pass | 14,376 | 9,632 | -33% | 1 | 1 | 0% | 2,249 | 3,356 | +49% | 0 | 0 | — |
case-18 | pass→pass | 12,395 | 11,481 | -7% | 1 | 1 | 0% | 1,885 | 3,626 | +92% | 0 | 0 | — |
case-20 | fail→pass | 17,785 | 11,709 | -34% | 1 | 1 | 0% | 2,505 | 3,578 | +43% | 0 | 0 | — |
case-21 | pass→pass | 21,221 | 13,624 | -36% | 1 | 1 | 0% | 3,004 | 4,117 | +37% | 0 | 0 | — |
case-22 | pass→pass | 7,467 | 11,388 | +53% | 1 | 1 | 0% | 1,026 | 3,346 | +226% | 0 | 0 | — |
case-23 | pass→fail | 20,514 | 19,964 | -3% | 1 | 1 | 0% | 3,230 | 5,094 | +58% | 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. 23 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 23 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.