Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Local iPolloWork Electron browser automation with CDP. Use when driving a local Electron dev app, browser_list, browser_snapshot, browser_eval, composer automation, or local UI smoke tests.
.claude/skills/devin-axis-browser-automation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-25 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -27% | 0% |
pnpm dev enables Electron CDP by default:
shIPOLLOWORK_ELECTRON_REMOTE_DEBUG_PORT=${IPOLLOWORK_ELECTRON_REMOTE_DEBUG_PORT:-9823}
The default browser URL for OpenCode browser tools is:
texthttp://127.0.0.1:9823
The app UI normally loads at:
texthttp://localhost:5173/
To use a different CDP port, launch with an override:
shIPOLLOWORK_ELECTRON_REMOTE_DEBUG_PORT=9830 pnpm dev
To disable Electron CDP for a run:
shIPOLLOWORK_ELECTRON_REMOTE_DEBUG_PORT=0 pnpm dev
Use a detached launch when the user wants the app running in the background:
shnohup pnpm dev > /var/folders/d9/xqhkvsp94rg0n0n523snqztm0000gn/T/opencode/ipollowork-dev.log 2>&1 &
Then wait for the CDP port:
shlsof -nP -iTCP:9823 -sTCP:LISTEN
browser_list using browser_url: "http://127.0.0.1:9823".iPolloWork target ID.browser_eval or browser_snapshot.[contenteditable="true"][data-lexical-editor="true"].Run task button.document.body.innerText or the current URL.Use this browser_eval pattern after selecting the iPolloWork target:
js(() => { const editor = document.querySelector('[contenteditable="true"][data-lexical-editor="true"]'); if (!editor) return { ok: false, reason: 'editor not found' }; editor.focus(); const data = new DataTransfer(); data.setData('text/plain', 'Say hello from the Electron browser test.'); editor.dispatchEvent(new ClipboardEvent('paste', { bubbles: true, cancelable: true, clipboardData: data, })); const run = Array.from(document.querySelectorAll('button')) .find((button) => button.innerText.trim() === 'Run task'); if (!run) return { ok: false, reason: 'Run task not found', inserted }; if (run.disabled) return { ok: false, reason: 'Run task disabled', inserted, text: editor.innerText }; run.click(); return { ok: true, inserted: true, text: editor.innerText }; })()
iPolloWork target at http://127.0.0.1:9823.Say hello from the Electron browser test. and the expected response is Hello from the Electron browser test.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-25 | fail→pass | 28,354 | 4,211 | -85% | 1 | 1 | 0% | 1,484 | 1,539 | +4% | 0 | 0 | — |
case-01 | fail→fail | 12,637 | 16,036 | +27% | 1 | 1 | 0% | 1,287 | 1,308 | +2% | 0 | 0 | — |
case-02 | fail→fail | 16,537 | 16,780 | +1% | 1 | 1 | 0% | 2,002 | 1,062 | -47% | 0 | 0 | — |
case-03 | fail→fail | 20,491 | 9,258 | -55% | 1 | 1 | 0% | 3,013 | 1,695 | -44% | 0 | 0 | — |
case-04 | fail→pass | 18,029 | 8,066 | -55% | 1 | 1 | 0% | 2,028 | 1,350 | -33% | 0 | 0 | — |
case-05 | fail→pass | 14,960 | 11,700 | -22% | 1 | 1 | 0% | 1,640 | 2,005 | +22% | 0 | 0 | — |
case-06 | pass→pass | 17,845 | 16,375 | -8% | 1 | 1 | 0% | 2,372 | 2,895 | +22% | 0 | 0 | — |
case-07 | fail→pass | 13,804 | 7,249 | -47% | 1 | 1 | 0% | 1,523 | 1,234 | -19% | 0 | 0 | — |
case-08 | fail→pass | 17,163 | 4,396 | -74% | 1 | 1 | 0% | 2,304 | 1,683 | -27% | 0 | 0 | — |
case-09 | fail→pass | 32,382 | 7,383 | -77% | 1 | 1 | 0% | 4,498 | 1,162 | -74% | 0 | 0 | — |
case-10 | fail→pass | 12,334 | 8,005 | -35% | 1 | 1 | 0% | 1,274 | 1,304 | +2% | 0 | 0 | — |
case-11 | fail→pass | 7,352 | 7,679 | +4% | 1 | 1 | 0% | 1,273 | 1,138 | -11% | 0 | 0 | — |
case-12 | fail→pass | 15,486 | 49,002 | +216% | 1 | 1 | 0% | 1,788 | 1,163 | -35% | 0 | 0 | — |
case-13 | pass→fail | 14,489 | 11,502 | -21% | 1 | 1 | 0% | 2,665 | 1,151 | -57% | 0 | 0 | — |
case-14 | pass→pass | 6,866 | 1,906 | -72% | 1 | 1 | 0% | 1,081 | 1,077 | -0% | 0 | 0 | — |
case-15 | fail→pass | 11,054 | 10,012 | -9% | 1 | 1 | 0% | 1,976 | 1,695 | -14% | 0 | 0 | — |
case-16 | fail→pass | 12,940 | 9,524 | -26% | 1 | 1 | 0% | 2,155 | 1,551 | -28% | 0 | 0 | — |
case-17 | fail→pass | 18,216 | 5,209 | -71% | 1 | 1 | 0% | 2,021 | 1,745 | -14% | 0 | 0 | — |
case-18 | fail→pass | 14,169 | 2,231 | -84% | 1 | 1 | 0% | 1,341 | 1,077 | -20% | 0 | 0 | — |
case-19 | fail→pass | 9,858 | 1,501 | -85% | 1 | 1 | 0% | 1,290 | 1,023 | -21% | 0 | 0 | — |
case-20 | pass→pass | 8,194 | 2,887 | -65% | 1 | 1 | 0% | 1,517 | 1,127 | -26% | 0 | 0 | — |
case-21 | fail→pass | 14,957 | 2,682 | -82% | 1 | 1 | 0% | 2,323 | 1,418 | -39% | 0 | 0 | — |
case-22 | fail→pass | 10,871 | 2,918 | -73% | 1 | 1 | 0% | 1,768 | 1,341 | -24% | 0 | 0 | — |
case-23 | fail→pass | 9,169 | 4,134 | -55% | 1 | 1 | 0% | 1,435 | 1,455 | +1% | 0 | 0 | — |
case-24 | fail→pass | 11,660 | 4,222 | -64% | 1 | 1 | 0% | 1,794 | 1,337 | -25% | 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 22 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 +68 percentage points is the difference between those two pass rates over the 22 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.