Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when writing Python code with `axllm` for audio input/output, OpenAI Responses audio mapping, realtime event folding, and generated package audio examples.
.claude/skills/ax-llm-ax-python-audio/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 97% | 79 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 0% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -41% | 0% |
This skill helps an agent write Python code with the generated Ax package axllm. Use the generated package API, examples, and manifests; do not import TypeScript-only APIs unless you are editing the TypeScript package.
axllm.API.md and axir-api.json.axir-capabilities.json.examples/.javascript-quickjs, python-pyodide.pythonimport os from axllm import ai llm = ai("openai", api_key=os.environ["OPENAI_API_KEY"])
ai, typesafe, AxAITypesafeClient, AxCancellationToken, AxAIServiceAbortedError, get_supported_ai_models, dict[str, str], Callable[[dict[str, str]], dict[str, str]], AIClient.owned_worker_factory, AxChatSession, closable generator, OpenAICompatibleClient, OpenAIResponsesClient, GoogleGeminiClient, AnthropicClient, AxUsageContext, AxUsageEvent, AxUsageObserver, set_usage_observer, AxRuntimeHooks, AxRateLimitInfo, AxRateLimiter, AxTracer, AxMeter, AxGlobals, set_rate_limiter, set_tracer, set_meter, AxBalancer, AxBalancerAdaptiveStrategy, AxBalancerStatsStore, AxInMemoryBalancerStatsStore, create_balancer_route_stats, update_balancer_route_stats, sample_balancer_route_health, MultiServiceRouter, ProviderRouterprovider-api examples only when the user explicitly has provider credentials available.no-key examples for deterministic local checks and provider request mapping.tools/*/skills/ into user packages.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 15,509 | 21,428 | +38% | 1 | 1 | 0% | 2,819 | 3,531 | +25% | 0 | 0 | — |
case-02 | fail→fail | 19,068 | 12,347 | -35% | 1 | 1 | 0% | 2,732 | 1,742 | -36% | 0 | 0 | — |
case-03 | fail→pass | 28,529 | 12,352 | -57% | 1 | 1 | 0% | 3,250 | 2,129 | -34% | 0 | 0 | — |
case-04 | pass→pass | 7,410 | 4,891 | -34% | 1 | 1 | 0% | 1,208 | 1,381 | +14% | 0 | 0 | — |
case-05 | pass→pass | 9,855 | 9,150 | -7% | 1 | 1 | 0% | 1,827 | 2,091 | +14% | 0 | 0 | — |
case-06 | pass→pass | 17,348 | 7,261 | -58% | 1 | 1 | 0% | 2,828 | 1,577 | -44% | 0 | 0 | — |
case-07 | pass→pass | 22,276 | 15,636 | -30% | 1 | 1 | 0% | 4,149 | 3,701 | -11% | 0 | 0 | — |
case-08 | fail→pass | 14,880 | 10,574 | -29% | 1 | 1 | 0% | 2,282 | 2,274 | -0% | 0 | 0 | — |
case-09 | fail→pass | 24,076 | 9,473 | -61% | 1 | 1 | 0% | 3,911 | 2,272 | -42% | 0 | 0 | — |
case-10 | pass→pass | 15,726 | 47,192 | +200% | 1 | 1 | 0% | 2,637 | 2,760 | +5% | 0 | 0 | — |
case-11 | fail→pass | 18,014 | 7,540 | -58% | 1 | 1 | 0% | 2,948 | 1,754 | -41% | 0 | 0 | — |
case-12 | fail→pass | 20,078 | 6,719 | -67% | 1 | 1 | 0% | 2,901 | 1,503 | -48% | 0 | 0 | — |
case-13 | fail→pass | 9,753 | 4,508 | -54% | 1 | 1 | 0% | 1,652 | 1,017 | -38% | 0 | 0 | — |
case-14 | fail→pass | 13,690 | 4,986 | -64% | 1 | 1 | 0% | 1,359 | 1,362 | +0% | 0 | 0 | — |
case-15 | fail→pass | 15,973 | 13,433 | -16% | 1 | 1 | 0% | 2,491 | 2,626 | +5% | 0 | 0 | — |
case-16 | fail→pass | 14,259 | 4,055 | -72% | 1 | 1 | 0% | 2,338 | 1,185 | -49% | 0 | 0 | — |
case-17 | fail→pass | 23,176 | 5,399 | -77% | 1 | 1 | 0% | 2,488 | 1,381 | -44% | 0 | 0 | — |
case-18 | fail→pass | 42,606 | 3,541 | -92% | 1 | 1 | 0% | 2,656 | 946 | -64% | 0 | 0 | — |
case-19 | pass→pass | 32,907 | 6,159 | -81% | 1 | 1 | 0% | 5,323 | 899 | -83% | 0 | 0 | — |
case-20 | pass→pass | 27,009 | 13,305 | -51% | 1 | 1 | 0% | 2,711 | 2,648 | -2% | 0 | 0 | — |
case-21 | pass→pass | 16,681 | 15,128 | -9% | 1 | 1 | 0% | 2,239 | 2,151 | -4% | 0 | 0 | — |
case-22 | fail→pass | 11,035 | 3,661 | -67% | 1 | 1 | 0% | 1,652 | 1,090 | -34% | 0 | 0 | — |
case-23 | fail→pass | 12,696 | 7,938 | -37% | 1 | 1 | 0% | 2,092 | 1,713 | -18% | 0 | 0 | — |
case-24 | fail→pass | 25,084 | 3,107 | -88% | 1 | 1 | 0% | 4,339 | 1,006 | -77% | 0 | 0 | — |
case-25 | fail→pass | 11,189 | 5,013 | -55% | 1 | 1 | 0% | 1,993 | 1,402 | -30% | 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. The headline lift of +64 percentage points is the difference between those two pass rates over the 25 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 | 9/9/2026 | +63% |
| gemini-3.6-flash | verified | 9/2/2026 | +83% |
| gemini-3.6-flash | verified | 8/26/2026 | +68% |
| gemini-3.6-flash | verified | 8/17/2026 | +59% |
| gemini-3.6-flash | verified | 8/10/2026 | +68% |
Other measured skills in the registry, with their headline benchmark lift.