Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when SynthLabs work must be driven through the visual app for verifier review, data preview, settings changes, DEEP mode, or other browser-led workflows.
.claude/skills/mkurman-synthlabs-ui-operator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 19% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 199% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 123% | 0% |
Use this skill when SynthLabs is best operated as an application instead of a backend route. Assume synthlabs-setup already confirmed a healthy instance before you begin. This skill does not provide dedicated zorai SynthLabs tools; it maps the work onto zorai's existing browser workflow.
Use this skill when:
Do not use this skill when:
synthlabs-setup confirmed SynthLabs is healthy.skills/operating/browser.md: start by opening a browser pane, read the DOM before interacting, use element discovery instead of guessing selectors, prefer text-based clicks when possible, and read the DOM again after navigation.open_canvas_browser, browser_read_dom, browser_get_elements, browser_click, and browser_type; do not imply SynthLabs-specific zorai tools exist.synthlabs-generation.synthlabs-curation.synthlabs-setup./api/sessions, /api/ai/generate, or /api/jobs, stop using the browser and switch to the backend-focused skill that owns that route.synthlabs-generation or synthlabs-curation when the UI-only phase is complete.GET /api/sessions, POST /api/sessions, or POST /api/ai/generate, switch to synthlabs-generation.GET /api/jobs, GET /api/jobs/:id, or a /api/jobs/* curation route, switch to synthlabs-curation./api/sessions or /api/jobs.synthlabs-setup first.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | fail→pass | 7,677 | 4,104 | -47% | 1 | 1 | 0% | 1,342 | 1,661 | +24% | 0 | 0 | — |
case-01 | fail→fail | 5,282 | 6,314 | +20% | 1 | 1 | 0% | 734 | 1,480 | +102% | 0 | 0 | — |
case-02 | fail→fail | 9,594 | 4,362 | -55% | 1 | 1 | 0% | 1,477 | 1,638 | +11% | 0 | 0 | — |
case-03 | fail→fail | 2,858 | 7,807 | +173% | 1 | 1 | 0% | 449 | 1,492 | +232% | 0 | 0 | — |
case-04 | fail→fail | 6,824 | 4,284 | -37% | 1 | 1 | 0% | 906 | 1,682 | +86% | 0 | 0 | — |
case-05 | pass→pass | 10,291 | 3,522 | -66% | 1 | 1 | 0% | 1,789 | 1,610 | -10% | 0 | 0 | — |
case-07 | fail→fail | 4,622 | 2,021 | -56% | 1 | 1 | 0% | 651 | 1,324 | +103% | 0 | 0 | — |
case-08 | pass→pass | 6,598 | 3,377 | -49% | 1 | 1 | 0% | 1,378 | 1,598 | +16% | 0 | 0 | — |
case-09 | pass→pass | 8,923 | 6,374 | -29% | 1 | 1 | 0% | 1,622 | 2,142 | +32% | 0 | 0 | — |
case-10 | fail→pass | 8,777 | 4,241 | -52% | 1 | 1 | 0% | 1,323 | 1,571 | +19% | 0 | 0 | — |
case-11 | fail→fail | 11,788 | 3,407 | -71% | 1 | 1 | 0% | 1,846 | 1,570 | -15% | 0 | 0 | — |
case-12 | pass→pass | 16,235 | 4,931 | -70% | 1 | 1 | 0% | 1,335 | 1,892 | +42% | 0 | 0 | — |
case-13 | fail→pass | 2,783 | 1,682 | -40% | 1 | 1 | 0% | 433 | 1,295 | +199% | 0 | 0 | — |
case-14 | pass→pass | 7,551 | 3,812 | -50% | 1 | 1 | 0% | 1,056 | 1,729 | +64% | 0 | 0 | — |
case-15 | fail→fail | 6,959 | 3,685 | -47% | 1 | 1 | 0% | 1,070 | 1,597 | +49% | 0 | 0 | — |
case-16 | pass→pass | 10,089 | 3,167 | -69% | 1 | 1 | 0% | 1,621 | 1,564 | -4% | 0 | 0 | — |
case-17 | pass→pass | 5,515 | 2,922 | -47% | 1 | 1 | 0% | 812 | 1,469 | +81% | 0 | 0 | — |
case-18 | fail→pass | 9,085 | 3,197 | -65% | 1 | 1 | 0% | 1,329 | 1,525 | +15% | 0 | 0 | — |
case-19 | pass→pass | 5,138 | 8,730 | +70% | 1 | 1 | 0% | 771 | 1,809 | +135% | 0 | 0 | — |
case-20 | fail→pass | 4,541 | 2,797 | -38% | 1 | 1 | 0% | 642 | 1,430 | +123% | 0 | 0 | — |
case-21 | fail→pass | 5,438 | 3,882 | -29% | 1 | 1 | 0% | 713 | 1,619 | +127% | 0 | 0 | — |
case-22 | fail→fail | 8,762 | 4,405 | -50% | 1 | 1 | 0% | 1,349 | 1,623 | +20% | 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 20 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 +27 percentage points is the difference between those two pass rates over the 20 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.