Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Manage device tier configs with gplay `device-tiers` — immutable device-targeting for tiered content delivery (create/view/list only, no update or delete). Use when defining or inspecting device tiers for tiered asset delivery, or scripting a config into CI.
.claude/skills/pollyglot-gplay-device-tiers/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 75% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -51% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -46% | 0% |
gplay device-tiers manages device tier configs, the app-scoped configuration that drives tiered content delivery: which devices fall into which tier, so Play can serve different assets to high-end and low-end devices. Shared conventions are in gplay-cli-usage. The whole namespace is [experimental].
A config is immutable: to change targeting, create a new one and point delivery at the new deviceTierConfigId; old configs stay. That is why create needs no --confirm. Outside the Edit lifecycle (no editId): gplay edits begin does not batch it.
bashgplay device-tiers create --file config.json # from a file cat config.json | gplay device-tiers create # from stdin (default when --file is omitted or "-") gplay device-tiers create --file config.json --dry-run # rehearse: validate the body, resolve the target, no HTTP
Body shape: gplay schema DeviceTierConfig expands its fields and enums offline (see gplay-cli-usage), or round-trip an existing config with view --output json.
bashgplay device-tiers list # newest first; --page-size / --page-token to page gplay device-tiers view <deviceTierConfigId> # one config by its server-assigned id gplay device-tiers view <id> --output json # DeviceTierConfig, verbatim
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 12,423 | 7,801 | -37% | 1 | 1 | 0% | 858 | 695 | -19% | 0 | 0 | — |
case-02 | fail→pass | 26,930 | 8,128 | -70% | 1 | 1 | 0% | 1,608 | 725 | -55% | 0 | 0 | — |
case-03 | fail→pass | 3,829 | 13,515 | +253% | 1 | 1 | 0% | 491 | 861 | +75% | 0 | 0 | — |
case-04 | fail→pass | 17,229 | 11,592 | -33% | 1 | 1 | 0% | 1,872 | 912 | -51% | 0 | 0 | — |
case-05 | pass→pass | 25,894 | 6,046 | -77% | 1 | 1 | 0% | 1,567 | 701 | -55% | 0 | 0 | — |
case-06 | fail→pass | 17,626 | 13,179 | -25% | 1 | 1 | 0% | 1,890 | 1,028 | -46% | 0 | 0 | — |
case-07 | pass→pass | 14,135 | 8,416 | -40% | 1 | 1 | 0% | 1,859 | 799 | -57% | 0 | 0 | — |
case-08 | pass→pass | 22,246 | 7,720 | -65% | 1 | 1 | 0% | 1,667 | 657 | -61% | 0 | 0 | — |
case-09 | pass→pass | 10,911 | 7,761 | -29% | 1 | 1 | 0% | 1,379 | 547 | -60% | 0 | 0 | — |
case-10 | fail→pass | 15,680 | 2,701 | -83% | 1 | 1 | 0% | 1,713 | 710 | -59% | 0 | 0 | — |
case-11 | pass→pass | 20,417 | 8,968 | -56% | 1 | 1 | 0% | 762 | 635 | -17% | 0 | 0 | — |
case-12 | pass→pass | 20,298 | 10,795 | -47% | 1 | 1 | 0% | 2,336 | 547 | -77% | 0 | 0 | — |
case-13 | fail→pass | 33,395 | 8,306 | -75% | 1 | 1 | 0% | 2,248 | 669 | -70% | 0 | 0 | — |
case-14 | fail→pass | 19,304 | 11,530 | -40% | 1 | 1 | 0% | 1,236 | 704 | -43% | 0 | 0 | — |
case-15 | pass→pass | 24,930 | 10,451 | -58% | 1 | 1 | 0% | 1,807 | 1,200 | -34% | 0 | 0 | — |
case-16 | pass→pass | 29,652 | 7,741 | -74% | 1 | 1 | 0% | 2,375 | 641 | -73% | 0 | 0 | — |
case-17 | fail→pass | 17,807 | 9,678 | -46% | 1 | 1 | 0% | 1,958 | 1,085 | -45% | 0 | 0 | — |
case-18 | pass→pass | 14,475 | 3,148 | -78% | 1 | 1 | 0% | 1,246 | 746 | -40% | 0 | 0 | — |
case-19 | pass→pass | 15,376 | 7,743 | -50% | 1 | 1 | 0% | 989 | 652 | -34% | 0 | 0 | — |
case-20 | fail→pass | 9,690 | 11,359 | +17% | 1 | 1 | 0% | 1,437 | 1,029 | -28% | 0 | 0 | — |
case-21 | fail→pass | 17,546 | 13,014 | -26% | 1 | 1 | 0% | 2,109 | 1,431 | -32% | 0 | 0 | — |
case-22 | pass→pass | 8,932 | 14,724 | +65% | 1 | 1 | 0% | 1,166 | 909 | -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. 22 cases were attempted. The headline lift of +50 percentage points is the difference between those two pass rates over the 22 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/24/2026 | +59% |
| gemini-3.6-flash | verified | 8/4/2026 | +59% |
Other measured skills in the registry, with their headline benchmark lift.