Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Google Agent Development Kit (ADK). Code-first Python toolkit for building, evaluating, and deploying AI agents. Multi-agent orchestration, tool integration, built-in evaluation, and deployment to Vertex AI.
.claude/skills/mkurman-google-adk/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -7% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -46% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -49% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -74% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -60% | 0% |
Google Agent Development Kit (ADK) is a code-first Python framework for building AI agents powered by Gemini. Supports tool integration, handoffs between agents, guardrails, multi-agent graphs, and deployment to Vertex AI Agent Builder for production serving.
bashuv pip install google-adk
pythonfrom google.adk.agents import Agent from google.adk.tools import FunctionTool def get_weather(location: str) -> str: return f"The weather in {location} is 22C and sunny." agent = Agent( name="weather_agent", model="gemini-2.0-flash", instruction="You are a helpful weather assistant.", tools=[FunctionTool(get_weather)], ) response = agent.run("What's the weather in Paris?") print(response.content)
pythonresearch = Agent(name="researcher", model="gemini-2.0-flash", ...) writer = Agent(name="writer", model="gemini-2.0-flash", ...) reviewer = Agent(name="reviewer", model="gemini-2.0-flash", ...) from google.adk.runners import Runner runner = Runner(agents=[research, writer, reviewer]) result = runner.run("Research and write about quantum computing.")
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 9,955 | 7,723 | -22% | 1 | 1 | 0% | 1,903 | 1,761 | -7% | 0 | 0 | — |
case-02 | fail→pass | 19,768 | 9,946 | -50% | 1 | 1 | 0% | 4,002 | 2,169 | -46% | 0 | 0 | — |
case-03 | fail→pass | 12,951 | 4,639 | -64% | 1 | 1 | 0% | 2,423 | 1,237 | -49% | 0 | 0 | — |
case-04 | pass→pass | 5,525 | 4,664 | -16% | 1 | 1 | 0% | 1,052 | 1,292 | +23% | 0 | 0 | — |
case-05 | pass→pass | 9,430 | 7,846 | -17% | 1 | 1 | 0% | 1,939 | 1,957 | +1% | 0 | 0 | — |
case-06 | pass→pass | 13,003 | 8,428 | -35% | 1 | 1 | 0% | 2,587 | 1,964 | -24% | 0 | 0 | — |
case-07 | fail→pass | 11,752 | 1,418 | -88% | 1 | 1 | 0% | 2,096 | 548 | -74% | 0 | 0 | — |
case-08 | pass→pass | 13,514 | 1,766 | -87% | 1 | 1 | 0% | 2,443 | 619 | -75% | 0 | 0 | — |
case-14 | fail→pass | 12,513 | 2,464 | -80% | 1 | 1 | 0% | 2,037 | 816 | -60% | 0 | 0 | — |
case-09 | fail→pass | 12,012 | 3,408 | -72% | 1 | 1 | 0% | 2,027 | 999 | -51% | 0 | 0 | — |
case-10 | pass→pass | 10,671 | 1,815 | -83% | 1 | 1 | 0% | 1,791 | 598 | -67% | 0 | 0 | — |
case-11 | fail→pass | 9,725 | 1,945 | -80% | 1 | 1 | 0% | 1,713 | 555 | -68% | 0 | 0 | — |
case-12 | fail→pass | 9,424 | 1,538 | -84% | 1 | 1 | 0% | 1,580 | 595 | -62% | 0 | 0 | — |
case-13 | pass→pass | 5,306 | 2,231 | -58% | 1 | 1 | 0% | 871 | 707 | -19% | 0 | 0 | — |
case-15 | pass→pass | 11,718 | 2,061 | -82% | 1 | 1 | 0% | 2,020 | 710 | -65% | 0 | 0 | — |
case-16 | pass→pass | 8,375 | 1,809 | -78% | 1 | 1 | 0% | 1,332 | 629 | -53% | 0 | 0 | — |
case-17 | pass→pass | 5,205 | 1,819 | -65% | 1 | 1 | 0% | 762 | 682 | -10% | 0 | 0 | — |
case-18 | fail→pass | 9,868 | 2,095 | -79% | 1 | 1 | 0% | 1,479 | 635 | -57% | 0 | 0 | — |
case-24 | fail→pass | 7,120 | 1,382 | -81% | 1 | 1 | 0% | 1,076 | 556 | -48% | 0 | 0 | — |
case-19 | fail→pass | 12,089 | 5,456 | -55% | 1 | 1 | 0% | 2,312 | 1,408 | -39% | 0 | 0 | — |
case-20 | pass→pass | 7,161 | 2,610 | -64% | 1 | 1 | 0% | 1,397 | 820 | -41% | 0 | 0 | — |
case-21 | pass→pass | 5,600 | 1,549 | -72% | 1 | 1 | 0% | 931 | 540 | -42% | 0 | 0 | — |
case-22 | pass→pass | 8,142 | 1,467 | -82% | 1 | 1 | 0% | 1,418 | 550 | -61% | 0 | 0 | — |
case-23 | fail→pass | 5,158 | 1,339 | -74% | 1 | 1 | 0% | 801 | 529 | -34% | 0 | 0 | — |
case-25 | pass→pass | 8,894 | 1,424 | -84% | 1 | 1 | 0% | 1,425 | 534 | -63% | 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 +48 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.
Other measured skills in the registry, with their headline benchmark lift.