Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Scans a website to find broken links (404s, 500s). Crawls internal pages, identifies broken outbound links, and reports source pages for easy fixing. Use this when the user asks to "check for broken links", "find 404s", "audit my links", or "is my site healthy".
.claude/skills/nowork-studio-broken-link-checker/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -46% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -24% | 0% |
You are a technical SEO specialist focused on website health and crawlability. Broken links hurt user experience and waste "crawl budget" from search engines.
Your goal is to identify broken links and provide a clear path to fixing them.
If the user didn't provide a URL, ask: > "Which website should I check for broken links?"
Once you have the URL, store it as $TARGET_URL.
Run the broken link checker script:
bashpython3 seo/broken-link-checker/scripts/checker.py --url "$TARGET_URL" --max-pages 50
Note: You can adjust --max-pages if the user wants a deeper scan.
The script will output a JSON report. Analyze the broken_links array:
source field).If no broken links are found, congratulate the user on a healthy site!
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 17,323 | 15,507 | -10% | 1 | 1 | 0% | 2,001 | 781 | -61% | 0 | 0 | — |
case-02 | fail→fail | 23,707 | 15,678 | -34% | 1 | 1 | 0% | 3,181 | 671 | -79% | 0 | 0 | — |
case-03 | fail→fail | 20,641 | 15,468 | -25% | 1 | 1 | 0% | 2,590 | 694 | -73% | 0 | 0 | — |
case-04 | pass→pass | 13,173 | 7,608 | -42% | 1 | 1 | 0% | 1,335 | 674 | -50% | 0 | 0 | — |
case-05 | fail→fail | 17,066 | 8,002 | -53% | 1 | 1 | 0% | 2,302 | 753 | -67% | 0 | 0 | — |
case-06 | pass→pass | 9,906 | 12,788 | +29% | 1 | 1 | 0% | 1,015 | 1,918 | +89% | 0 | 0 | — |
case-07 | pass→pass | 11,933 | 7,975 | -33% | 1 | 1 | 0% | 1,049 | 929 | -11% | 0 | 0 | — |
case-08 | pass→pass | 15,975 | 11,944 | -25% | 1 | 1 | 0% | 1,916 | 1,631 | -15% | 0 | 0 | — |
case-09 | pass→pass | 15,194 | 10,882 | -28% | 1 | 1 | 0% | 1,569 | 1,428 | -9% | 0 | 0 | — |
case-10 | fail→pass | 12,568 | 8,181 | -35% | 1 | 1 | 0% | 1,222 | 934 | -24% | 0 | 0 | — |
case-11 | fail→pass | 22,305 | 11,197 | -50% | 1 | 1 | 0% | 2,718 | 1,461 | -46% | 0 | 0 | — |
case-12 | pass→pass | 12,259 | 7,073 | -42% | 1 | 1 | 0% | 1,248 | 768 | -38% | 0 | 0 | — |
case-13 | fail→pass | 15,001 | 7,399 | -51% | 1 | 1 | 0% | 1,689 | 898 | -47% | 0 | 0 | — |
case-14 | fail→pass | 12,324 | 13,754 | +12% | 1 | 1 | 0% | 1,233 | 637 | -48% | 0 | 0 | — |
case-15 | fail→pass | 13,979 | 9,323 | -33% | 1 | 1 | 0% | 1,591 | 1,217 | -24% | 0 | 0 | — |
case-16 | pass→pass | 14,053 | 9,103 | -35% | 1 | 1 | 0% | 1,428 | 1,064 | -25% | 0 | 0 | — |
case-17 | pass→pass | 17,449 | 12,615 | -28% | 1 | 1 | 0% | 2,090 | 1,821 | -13% | 0 | 0 | — |
case-18 | fail→pass | 14,514 | 7,871 | -46% | 1 | 1 | 0% | 1,754 | 930 | -47% | 0 | 0 | — |
case-19 | fail→pass | 18,966 | 6,462 | -66% | 1 | 1 | 0% | 2,406 | 673 | -72% | 0 | 0 | — |
case-20 | pass→pass | 14,846 | 13,032 | -12% | 1 | 1 | 0% | 2,110 | 1,993 | -6% | 0 | 0 | — |
case-21 | pass→pass | 12,129 | 11,904 | -2% | 1 | 1 | 0% | 1,231 | 1,669 | +36% | 0 | 0 | — |
case-22 | pass→fail | 21,934 | 30,405 | +39% | 1 | 1 | 0% | 2,955 | 2,820 | -5% | 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, and 18 counted toward the lift figure. The other 4 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +27 percentage points is the difference between those two pass rates over the 18 comparable cases. 1 case got worse with the skill loaded, and it is 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/12/2026 | +22% |
Other measured skills in the registry, with their headline benchmark lift.