Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Launch and control standalone Chrome in a Daytona sandbox via CDP. Use for web sign-in, OAuth, Den Web setup, browser-only flows, or when the app should not be driven through Electron CDP.
.claude/skills/devin-axis-daytona-chrome-cdp/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -26% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -2% | 0% |
Use this skill when a Daytona sandbox needs a normal Chrome/Chromium browser in the XFCE display, separate from the Electron app. This is useful for Den Web sign-in, OAuth provider setup, browser-only admin flows, and checking what a user would see in a regular browser.
Run inside the Electron or server sandbox:
bashdaytona exec "$SANDBOX" -- "bash -lc 'mkdir -p /tmp/daytona-chrome-profile; DISPLAY=:99 nohup chromium --no-sandbox --disable-dev-shm-usage --remote-debugging-address=0.0.0.0 --remote-debugging-port=9222 --user-data-dir=/tmp/daytona-chrome-profile \"$DEN_WEB_URL\" >/tmp/daytona-chrome.log 2>&1 &'"
If chromium is missing, try google-chrome, google-chrome-stable, or install Chromium in the sandbox only when needed.
bashCHROME_CDP_URL=$(daytona preview-url "$SANDBOX" -p 9222 2>/dev/null | grep -v '^time=')
Verify the target directly:
bashcurl -fsS "$CHROME_CDP_URL/json/list"
The coded flow must assert that navigator.userAgent contains Chrome/ or Chromium/ and does not contain Electron/.
Create or reuse an evals/flows/*.flow.mjs flow and set its cdpTarget to a stable title or URL fragment for the standalone page. Drive the visible flow with ctx.clickText, ctx.fill, ctx.waitFor, and ctx.screenshot, then run:
bashpnpm evals --flow <flow-id> --cdp-url "$CHROME_CDP_URL"
Validate with the fraimz loop. Do not assume navigation or sign-in worked until the post-action URL, visible assertion, and frame prove it.
bashdaytona exec "$SANDBOX" -- "bash -lc 'pkill -f \"chromium.*remote-debugging-port=9222\" || pkill -f \"chrome.*remote-debugging-port=9222\" || true'"
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 15,262 | 10,141 | -34% | 1 | 1 | 0% | 1,923 | 1,642 | -15% | 0 | 0 | — |
case-02 | fail→pass | 30,038 | 14,090 | -53% | 1 | 1 | 0% | 4,793 | 2,484 | -48% | 0 | 0 | — |
case-03 | fail→pass | 20,430 | 12,516 | -39% | 1 | 1 | 0% | 2,958 | 2,063 | -30% | 0 | 0 | — |
case-04 | pass→pass | 13,371 | 10,413 | -22% | 1 | 1 | 0% | 1,384 | 1,430 | +3% | 0 | 0 | — |
case-05 | pass→pass | 16,910 | 8,366 | -51% | 1 | 1 | 0% | 2,146 | 2,237 | +4% | 0 | 0 | — |
case-06 | pass→pass | 13,697 | 15,353 | +12% | 1 | 1 | 0% | 2,546 | 2,732 | +7% | 0 | 0 | — |
case-07 | pass→pass | 11,043 | 11,567 | +5% | 1 | 1 | 0% | 2,041 | 1,846 | -10% | 0 | 0 | — |
case-08 | fail→pass | 18,566 | 8,574 | -54% | 1 | 1 | 0% | 1,694 | 1,261 | -26% | 0 | 0 | — |
case-09 | fail→pass | 22,258 | 7,440 | -67% | 1 | 1 | 0% | 1,065 | 1,040 | -2% | 0 | 0 | — |
case-10 | fail→pass | 8,466 | 7,813 | -8% | 1 | 1 | 0% | 1,605 | 1,200 | -25% | 0 | 0 | — |
case-11 | fail→pass | 5,816 | 8,117 | +40% | 1 | 1 | 0% | 1,059 | 1,031 | -3% | 0 | 0 | — |
case-12 | pass→pass | 12,325 | 3,905 | -68% | 1 | 1 | 0% | 2,192 | 1,290 | -41% | 0 | 0 | — |
case-13 | fail→pass | 22,724 | 3,207 | -86% | 1 | 1 | 0% | 4,007 | 1,046 | -74% | 0 | 0 | — |
case-14 | fail→pass | 7,262 | 2,132 | -71% | 1 | 1 | 0% | 1,266 | 1,017 | -20% | 0 | 0 | — |
case-15 | pass→pass | 14,317 | 10,686 | -25% | 1 | 1 | 0% | 2,309 | 1,609 | -30% | 0 | 0 | — |
case-16 | pass→pass | 18,971 | 7,697 | -59% | 1 | 1 | 0% | 2,967 | 1,950 | -34% | 0 | 0 | — |
case-17 | fail→pass | 17,668 | 3,436 | -81% | 1 | 1 | 0% | 976 | 1,131 | +16% | 0 | 0 | — |
case-18 | fail→pass | 12,880 | 3,909 | -70% | 1 | 1 | 0% | 1,979 | 1,316 | -34% | 0 | 0 | — |
case-19 | pass→pass | 7,992 | 4,409 | -45% | 1 | 1 | 0% | 1,406 | 1,343 | -4% | 0 | 0 | — |
case-20 | fail→pass | 12,834 | 6,749 | -47% | 1 | 1 | 0% | 2,082 | 1,685 | -19% | 0 | 0 | — |
case-21 | fail→pass | 12,799 | 2,627 | -79% | 1 | 1 | 0% | 1,997 | 1,078 | -46% | 0 | 0 | — |
case-22 | fail→pass | 10,534 | 6,319 | -40% | 1 | 1 | 0% | 1,944 | 1,652 | -15% | 0 | 0 | — |
case-23 | pass→pass | 8,602 | 2,314 | -73% | 1 | 1 | 0% | 1,346 | 927 | -31% | 0 | 0 | — |
case-24 | pass→pass | 4,145 | 2,463 | -41% | 1 | 1 | 0% | 593 | 914 | +54% | 0 | 0 | — |
case-25 | fail→pass | 13,526 | 6,912 | -49% | 1 | 1 | 0% | 2,101 | 1,778 | -15% | 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. 25 cases were attempted, and 23 counted toward the lift figure. The other 2 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 +60 percentage points is the difference between those two pass rates over the 23 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.