Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Allows to interact with web pages by performing actions such as clicking buttons, filling out forms, and navigating links. It works by remote controlling Google Chrome or Chromium browsers using the Chrome DevTools Protocol (CDP). When Claude needs to browse the web, it can use this skill to do so.
.claude/skills/dicklesworthstone-web-browser/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -44% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -60% | 0% |
Minimal CDP tools for collaborative site exploration.
bash./scripts/start.js # Fresh profile ./scripts/start.js --profile # Copy your profile (cookies, logins)
Start Chrome on :9222 with remote debugging.
bash./scripts/nav.js https://example.com ./scripts/nav.js https://example.com --new
Navigate current tab or open new tab.
bash./scripts/eval.js 'document.title' ./scripts/eval.js 'document.querySelectorAll("a").length' ./scripts/eval.js 'JSON.stringify(Array.from(document.querySelectorAll("a")).map(a => ({ text: a.textContent.trim(), href: a.href })).filter(link => !link.href.startsWith("https://")))'
Execute JavaScript in active tab (async context). Be careful with string escaping, best to use single quotes.
bash./scripts/screenshot.js
Screenshot current viewport, returns temp file path
bash./scripts/pick.js "Click the submit button"
Interactive element picker. Click to select, Cmd/Ctrl+Click for multi-select, Enter to finish.
bash./scripts/dismiss-cookies.js # Accept cookies ./scripts/dismiss-cookies.js --reject # Reject cookies (where possible)
Automatically dismisses EU cookie consent dialogs.
Run after navigating to a page:
bash./scripts/nav.js https://example.com && ./scripts/dismiss-cookies.js
Automatically started by start.js and writes JSONL logs to:
~/.cache/agent-web/logs/YYYY-MM-DD/<targetId>.jsonlManually start:
bash./scripts/watch.js
Tail latest log:
bash./scripts/logs-tail.js # dump current log and exit ./scripts/logs-tail.js --follow # keep following
Summarize network responses:
bash./scripts/net-summary.js
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | fail→pass | 8,834 | 1,970 | -78% | 1 | 1 | 0% | 1,537 | 796 | -48% | 0 | 0 | — |
case-01 | fail→fail | 10,745 | 12,645 | +18% | 1 | 1 | 0% | 1,904 | 769 | -60% | 0 | 0 | — |
case-02 | fail→fail | 6,652 | 6,044 | -9% | 1 | 1 | 0% | 906 | 846 | -7% | 0 | 0 | — |
case-03 | fail→fail | 7,715 | 5,619 | -27% | 1 | 1 | 0% | 1,003 | 699 | -30% | 0 | 0 | — |
case-05 | pass→pass | 13,237 | 3,626 | -73% | 1 | 1 | 0% | 2,281 | 847 | -63% | 0 | 0 | — |
case-06 | fail→pass | 9,702 | 2,308 | -76% | 1 | 1 | 0% | 1,580 | 881 | -44% | 0 | 0 | — |
case-07 | fail→pass | 9,135 | 3,560 | -61% | 1 | 1 | 0% | 1,460 | 1,026 | -30% | 0 | 0 | — |
case-08 | fail→pass | 8,211 | 2,005 | -76% | 1 | 1 | 0% | 1,520 | 812 | -47% | 0 | 0 | — |
case-09 | fail→pass | 13,337 | 1,715 | -87% | 1 | 1 | 0% | 2,013 | 810 | -60% | 0 | 0 | — |
case-10 | fail→pass | 18,915 | 2,893 | -85% | 1 | 1 | 0% | 3,031 | 915 | -70% | 0 | 0 | — |
case-11 | fail→fail | 17,815 | 1,489 | -92% | 1 | 1 | 0% | 3,110 | 733 | -76% | 0 | 0 | — |
case-12 | fail→pass | 9,706 | 1,751 | -82% | 1 | 1 | 0% | 1,687 | 763 | -55% | 0 | 0 | — |
case-13 | fail→pass | 7,848 | 1,730 | -78% | 1 | 1 | 0% | 1,211 | 738 | -39% | 0 | 0 | — |
case-18 | pass→pass | 6,808 | 1,962 | -71% | 1 | 1 | 0% | 1,098 | 827 | -25% | 0 | 0 | — |
case-14 | fail→fail | 12,359 | 1,959 | -84% | 1 | 1 | 0% | 1,967 | 744 | -62% | 0 | 0 | — |
case-15 | fail→pass | 14,277 | 1,954 | -86% | 1 | 1 | 0% | 2,374 | 794 | -67% | 0 | 0 | — |
case-16 | fail→pass | 10,569 | 2,026 | -81% | 1 | 1 | 0% | 1,815 | 802 | -56% | 0 | 0 | — |
case-17 | fail→pass | 5,231 | 2,235 | -57% | 1 | 1 | 0% | 823 | 816 | -1% | 0 | 0 | — |
case-19 | fail→pass | 10,893 | 4,095 | -62% | 1 | 1 | 0% | 1,773 | 1,264 | -29% | 0 | 0 | — |
case-20 | fail→pass | 10,579 | 2,187 | -79% | 1 | 1 | 0% | 1,720 | 874 | -49% | 0 | 0 | — |
case-21 | pass→pass | 10,436 | 1,362 | -87% | 1 | 1 | 0% | 1,458 | 691 | -53% | 0 | 0 | — |
case-22 | pass→pass | 10,574 | 5,195 | -51% | 1 | 1 | 0% | 1,886 | 1,448 | -23% | 0 | 0 | — |
case-23 | pass→pass | 10,988 | 11,467 | +4% | 1 | 1 | 0% | 2,144 | 2,665 | +24% | 0 | 0 | — |
case-24 | pass→pass | 13,800 | 7,531 | -45% | 1 | 1 | 0% | 2,370 | 1,850 | -22% | 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. 24 cases were attempted, and 21 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 +54 percentage points is the difference between those two pass rates over the 21 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.