Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Automated web application testing using browser automation. Use when asked to test a web app, verify UI behavior, check for regressions, or automate browser interactions such as clicking, form filling, navigation, and screenshot capture.
.claude/skills/kxiandaoyan-webapp-testing/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-15 | ✗→✓ | ▲ Improved | -66% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -63% | 0% |
Test web applications through browser automation.
bash# Launch browser and navigate exec: agent-browser open https://example.com # Take screenshot exec: agent-browser screenshot --output /data/screenshot.png # Click element exec: agent-browser click --selector "#submit-btn" # Fill form exec: agent-browser fill --selector "input[name=email]" --value "test@example.com" # Assert text present exec: agent-browser assert-text --text "Success"
open URL → fill username → fill password → click submit → assert dashboard visiblefill fields → click submit → assert success message or error handlingclick nav link → assert URL changed → assert page content| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | pass→pass | 8,942 | 6,503 | -27% | 1 | 1 | 0% | 1,702 | 1,570 | -8% | 0 | 0 | — |
case-06 | pass→pass | 19,292 | 14,703 | -24% | 1 | 1 | 0% | 3,047 | 2,507 | -18% | 0 | 0 | — |
case-01 | fail→fail | 9,356 | 7,795 | -17% | 1 | 1 | 0% | 1,841 | 1,004 | -45% | 0 | 0 | — |
case-02 | fail→fail | 12,145 | 32,235 | +165% | 1 | 1 | 0% | 2,440 | 497 | -80% | 0 | 0 | — |
case-03 | fail→fail | 12,040 | 3,910 | -68% | 1 | 1 | 0% | 2,323 | 576 | -75% | 0 | 0 | — |
case-04 | pass→pass | 21,676 | 17,420 | -20% | 1 | 1 | 0% | 3,783 | 3,332 | -12% | 0 | 0 | — |
case-07 | fail→fail | 7,744 | 7,591 | -2% | 1 | 1 | 0% | 1,511 | 625 | -59% | 0 | 0 | — |
case-08 | pass→pass | 11,241 | 4,749 | -58% | 1 | 1 | 0% | 1,817 | 1,140 | -37% | 0 | 0 | — |
case-09 | pass→pass | 16,149 | 9,534 | -41% | 1 | 1 | 0% | 2,613 | 2,026 | -22% | 0 | 0 | — |
case-10 | fail→fail | 6,603 | 4,878 | -26% | 1 | 1 | 0% | 1,256 | 526 | -58% | 0 | 0 | — |
case-15 | fail→pass | 8,719 | 1,592 | -82% | 1 | 1 | 0% | 1,558 | 533 | -66% | 0 | 0 | — |
case-11 | fail→pass | 8,781 | 2,581 | -71% | 1 | 1 | 0% | 1,595 | 732 | -54% | 0 | 0 | — |
case-12 | fail→fail | 7,782 | 5,499 | -29% | 1 | 1 | 0% | 1,603 | 658 | -59% | 0 | 0 | — |
case-13 | fail→pass | 9,936 | 6,755 | -32% | 1 | 1 | 0% | 2,172 | 1,527 | -30% | 0 | 0 | — |
case-14 | pass→pass | 13,278 | 4,232 | -68% | 1 | 1 | 0% | 2,196 | 981 | -55% | 0 | 0 | — |
case-16 | fail→pass | 9,542 | 3,303 | -65% | 1 | 1 | 0% | 1,636 | 825 | -50% | 0 | 0 | — |
case-17 | fail→fail | 8,642 | 4,831 | -44% | 1 | 1 | 0% | 1,979 | 530 | -73% | 0 | 0 | — |
case-18 | fail→pass | 7,391 | 1,583 | -79% | 1 | 1 | 0% | 1,420 | 530 | -63% | 0 | 0 | — |
case-19 | fail→pass | 10,999 | 3,351 | -70% | 1 | 1 | 0% | 1,687 | 779 | -54% | 0 | 0 | — |
case-20 | fail→fail | 10,444 | 4,348 | -58% | 1 | 1 | 0% | 2,062 | 466 | -77% | 0 | 0 | — |
case-21 | fail→pass | 12,378 | 4,834 | -61% | 1 | 1 | 0% | 2,493 | 1,277 | -49% | 0 | 0 | — |
case-22 | fail→pass | 7,633 | 4,894 | -36% | 1 | 1 | 0% | 1,579 | 739 | -53% | 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 16 counted toward the lift figure. The other 6 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 16 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.
Other measured skills in the registry, with their headline benchmark lift.