Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Search Google via Bright Data SERP API. Returns structured JSON results with title, link, and description. Requires BRIGHTDATA_API_KEY and BRIGHTDATA_UNLOCKER_ZONE environment variables.
.claude/skills/davila7-search/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -49% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -65% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -72% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -48% | 0% |
Search Google and get structured JSON results using Bright Data's SERP API.
1. Get your API Key: Get a key from Bright Data Dashboard.
2. Create a Web Unlocker zone: Create a zone at brightdata.com/cp by clicking "Add" (top-right), selecting "Unlocker zone".
3. Set environment variables:
bashexport BRIGHTDATA_API_KEY="your-api-key" export BRIGHTDATA_UNLOCKER_ZONE="your-zone-name"
bashbash scripts/search.sh "query" [cursor]
Parameters:
query (required): Search termcursor (optional): Page number for pagination (0-indexed, default: 0)Examples:
bash# Basic search bash scripts/search.sh "climate change" # Get page 2 of results bash scripts/search.sh "climate change" 1
Returns JSON with structured organic array:
json{ "organic": [ { "link": "https://example.com/article", "title": "Article Title", "description": "Brief description of the page..." } ] }
curl - For API requestsjq - For JSON processing| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | pass→pass | 6,633 | 3,247 | -51% | 1 | 1 | 0% | 1,067 | 576 | -46% | 0 | 0 | — |
case-01 | fail→fail | 2,622 | 5,348 | +104% | 1 | 1 | 0% | 290 | 469 | +62% | 0 | 0 | — |
case-02 | fail→pass | 7,566 | 2,024 | -73% | 1 | 1 | 0% | 1,385 | 499 | -64% | 0 | 0 | — |
case-03 | fail→fail | 4,608 | 8,248 | +79% | 1 | 1 | 0% | 179 | 460 | +157% | 0 | 0 | — |
case-04 | fail→fail | 10,178 | 5,539 | -46% | 1 | 1 | 0% | 1,939 | 537 | -72% | 0 | 0 | — |
case-05 | fail→pass | 8,369 | 2,423 | -71% | 1 | 1 | 0% | 1,396 | 710 | -49% | 0 | 0 | — |
case-07 | fail→pass | 8,028 | 1,297 | -84% | 1 | 1 | 0% | 1,374 | 487 | -65% | 0 | 0 | — |
case-08 | fail→pass | 13,417 | 2,174 | -84% | 1 | 1 | 0% | 2,306 | 654 | -72% | 0 | 0 | — |
case-09 | pass→pass | 4,567 | 1,334 | -71% | 1 | 1 | 0% | 747 | 496 | -34% | 0 | 0 | — |
case-10 | pass→pass | 12,509 | 1,446 | -88% | 1 | 1 | 0% | 2,034 | 529 | -74% | 0 | 0 | — |
case-11 | fail→pass | 5,865 | 2,019 | -66% | 1 | 1 | 0% | 937 | 483 | -48% | 0 | 0 | — |
case-12 | pass→pass | 2,256 | 1,283 | -43% | 1 | 1 | 0% | 306 | 494 | +61% | 0 | 0 | — |
case-13 | fail→pass | 6,616 | 1,166 | -82% | 1 | 1 | 0% | 1,111 | 507 | -54% | 0 | 0 | — |
case-14 | fail→pass | 5,795 | 1,239 | -79% | 1 | 1 | 0% | 768 | 444 | -42% | 0 | 0 | — |
case-15 | fail→pass | 10,022 | 2,088 | -79% | 1 | 1 | 0% | 1,756 | 652 | -63% | 0 | 0 | — |
case-16 | pass→pass | 3,930 | 2,096 | -47% | 1 | 1 | 0% | 615 | 627 | +2% | 0 | 0 | — |
case-17 | pass→pass | 6,338 | 2,029 | -68% | 1 | 1 | 0% | 961 | 504 | -48% | 0 | 0 | — |
case-18 | pass→pass | 3,661 | 1,698 | -54% | 1 | 1 | 0% | 515 | 531 | +3% | 0 | 0 | — |
case-19 | pass→pass | 3,370 | 1,225 | -64% | 1 | 1 | 0% | 438 | 449 | +3% | 0 | 0 | — |
case-20 | pass→pass | 15,868 | 9,773 | -38% | 1 | 1 | 0% | 2,873 | 2,022 | -30% | 0 | 0 | — |
case-21 | pass→pass | 9,359 | 8,124 | -13% | 1 | 1 | 0% | 1,745 | 1,796 | +3% | 0 | 0 | — |
case-22 | pass→pass | 8,969 | 4,955 | -45% | 1 | 1 | 0% | 1,683 | 1,209 | -28% | 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 19 counted toward the lift figure. The other 3 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 +36 percentage points is the difference between those two pass rates over the 19 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.