genkit-ai/skills

buttons), not just prose. The whole server-side integration is the `a2ui()` model middleware in an agent's (or `ai.generate`'s) `use` list; the Flutter client renders surfaces with the [`genui`](https://pub.dev/packages/genui) package plus

middleware in an agent's (or `ai.generate`'s) `use` list; the Flutter client renders surfaces with the [`genui`](https://pub.dev/packages/genui) package plus the helpers in `package:genkit_a2ui/client.dart`. Dart specific: you must

renders surfaces with the [`genui`](https://pub.dev/packages/genui) package plus

curl -sL cli.genkit.dev | bash # Native CLI

curl -sL cli.genkit.dev | bash # Native CLI

**Primary pattern (default):** prefix `genkit start --` to your normal run command. This collects telemetry from any Genkit code your program runs, whether triggered from the dev UI, your own web server/web UI, or a plain script. Starts the

## External Dependencies Whenever you define schemas mapping inside of Tools, Flows, and Prompts, you must use the [schemantic](https://pub.dev/packages/schemantic) library.

Whenever you define schemas mapping inside of Tools, Flows, and Prompts, you must use the [schemantic](https://pub.dev/packages/schemantic) library.

The `genkit_a2ui` package brings > [A2UI](https://a2ui.org/), a transport-agnostic, JSON-based streaming UI

> [A2UI](https://a2ui.org/), a transport-agnostic, JSON-based streaming UI

To render surfaces you also need a renderer. The Flutter renderer for A2UI is [`genui`](https://pub.dev/packages/genui). On the **client** Flutter app, add it

[`genui`](https://pub.dev/packages/genui). On the **client** Flutter app, add it

// remoteAgent returns an AgentApi<State> (not a `RemoteAgent`). final agent = remoteAgent(   url: 'http://localhost:8080/api/uiAgent',

final agent = remoteAgent(   url: 'http://localhost:8080/api/uiAgent',   getSnapshotUrl: 'http://localhost:8080/api/uiAgent/getSnapshot',

url: 'http://localhost:8080/api/uiAgent',

getSnapshotUrl: 'http://localhost:8080/api/uiAgent/getSnapshot',

abortUrl: 'http://localhost:8080/api/uiAgent/abort',

port,   );   print('Agents API server running on http://localhost:${server.port}');

);   print('Agents API server running on http://localhost:${server.port}'); }

print('Agents API server running on http://localhost:${server.port}');

final weather = remoteAgent(url: 'http://localhost:8080/api/weatherAgent');

final agent = remoteAgent(   url: 'https://your-backend.example.com/api/weatherAgent',

url: 'https://your-backend.example.com/api/weatherAgent',

class _ChatViewState extends State<ChatView> {   late final AgentApi _agent = remoteAgent(     url: 'http://localhost:8080/api/weatherAgent',

late final AgentApi _agent = remoteAgent(     url: 'http://localhost:8080/api/weatherAgent',   );

url: 'http://localhost:8080/api/weatherAgent',

Picoschema follows the [spec](https://genkit.dev/docs/dotprompt/#picoschema),

final stringAction = defineRemoteAction(   url: 'http://localhost:3400/my-flow',

url: 'http://localhost:3400/my-flow',

final streamAction = defineRemoteAction(   url: 'http://localhost:3400/stream-story',

url: 'http://localhost:3400/stream-story',

try {     const result = await runFlow({       url: 'http://127.0.0.1:3400/helloFlow', // Replace with your deployed flow's URL

const result = await runFlow({       url: 'http://127.0.0.1:3400/helloFlow', // Replace with your deployed flow's URL       input: { name: 'Genkit User' },

url: 'http://127.0.0.1:3400/helloFlow', // Replace with your deployed flow's URL

try {     const result = streamFlow({       url: 'http://127.0.0.1:3400/helloFlow', // Replace with your deployed flow's URL

const result = streamFlow({       url: 'http://127.0.0.1:3400/helloFlow', // Replace with your deployed flow's URL       input: { name: 'Streaming User' },

url: 'http://127.0.0.1:3400/helloFlow', // Replace with your deployed flow's URL

// A Media/Image part   final mediaPart = MediaPart(     media: Media(url: 'https://...', contentType: 'image/png'),

final mediaPart = MediaPart(     media: Media(url: 'https://...', contentType: 'image/png'),     metadata: {'foo': 'bar'},

media: Media(url: 'https://...', contentType: 'image/png'),

[Credential-related source omitted]

[Credential-related source omitted]

name: 'my-host',     mcpServers: {       'remote': McpServerConfig(url: Uri.parse('https://mcp.example.com/mcp')),

mcpServers: {       'remote': McpServerConfig(url: Uri.parse('https://mcp.example.com/mcp')),     },

'remote': McpServerConfig(url: Uri.parse('https://mcp.example.com/mcp')),

>   port: 3000, >   allowedHosts: ['mcp.example.com'], >   allowedOrigins: ['https://app.example.com'],

>   allowedHosts: ['mcp.example.com'], >   allowedOrigins: ['https://app.example.com'], > );

>   allowedOrigins: ['https://app.example.com'],

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

baseUrl: 'https://api.groq.com/openai/v1',

HTTP serving lives in core `package:genkit/io.dart` (`GenkitRouter`, plain `dart:io`, no web framework). `genkit_shelf` is only an adapter that mounts a `GenkitRouter` into a [shelf](https://pub.dev/packages/shelf) app.

`GenkitRouter` into a [shelf](https://pub.dev/packages/shelf) app.

await genkit.serve(     port: 8080, // default: $PORT, then 3400. Host defaults to 0.0.0.0.     cors: const CorsOptions(allowedOrigins: ['https://myapp.dev']), // default: no CORS

port: 8080, // default: $PORT, then 3400. Host defaults to 0.0.0.0.     cors: const CorsOptions(allowedOrigins: ['https://myapp.dev']), // default: no CORS   );

cors: const CorsOptions(allowedOrigins: ['https://myapp.dev']), // default: no CORS

runtimes it is automatic. The account needs `roles/aiplatform.user`. - **No API key.** For API-key access use `genkit_google_genai` (`googleAI()`). - `authClient:` takes your own authenticated `http.Client`; the plugin never

- **No API key.** For API-key access use `genkit_google_genai` (`googleAI()`). - `authClient:` takes your own authenticated `http.Client`; the plugin never   closes it.

- `authClient:` takes your own authenticated `http.Client`; the plugin never

curl -sL cli.genkit.dev | bash

curl -sL cli.genkit.dev | bash

**Primary pattern (default):** prefix `genkit start --` to your normal run command. This collects telemetry from any Genkit code your program runs, whether triggered from the dev UI, your own web server/web UI, or a plain script. Starts the

The A2UI plugin (`github.com/firebase/genkit/go/plugins/a2ui/exp`, imported as `a2uix`) brings [A2UI](https://a2ui.org/), a transport-agnostic, JSON-based

`a2uix`) brings [A2UI](https://a2ui.org/), a transport-agnostic, JSON-based

`actionToMessage`). - **Dart/Flutter**: rendering with the   [`genui`](https://pub.dev/packages/genui) package plus the

- **Dart/Flutter**: rendering with the   [`genui`](https://pub.dev/packages/genui) package plus the   `package:genkit_a2ui/client.dart` helpers.

[`genui`](https://pub.dev/packages/genui) package plus the

curl -N -X POST 'http://localhost:8080/api/uiAgent?stream=true' \

weatherCatalog := a2uix.BasicCatalog() weatherCatalog.ID = "https://my-app.org/catalogs/weather.json"

weatherCatalog.ID = "https://my-app.org/catalogs/weather.json"

# One-shot turn: curl -s localhost:8080/api/weatherAgent \

curl -s localhost:8080/api/weatherAgent \

# Resume a server-managed session: curl -s localhost:8080/api/weatherAgent \

curl -s localhost:8080/api/weatherAgent \

researcher := genkitx.DefineRemoteAgent(g, "researcher", "https://research.internal/agents/researcher",

curl -X POST http://localhost:8080/jokeFlow \

curl -N -X POST http://localhost:8080/streamingJokeFlow \

curl -sL cli.genkit.dev | bash

curl -sL cli.genkit.dev | bash

This launches: - Your app (with tracing enabled) - The Developer UI at `http://localhost:4000`

- Your app (with tracing enabled) - The Developer UI at `http://localhost:4000` - A telemetry API at `http://localhost:4033`

- The Developer UI at `http://localhost:4000`

- A telemetry API at `http://localhost:4033`

[Credential-related source omitted]

Model names follow the format `googleai/<model-id>`. Look up the latest model IDs at https://ai.google.dev/gemini-api/docs/models.

[Credential-related source omitted]

Model names follow the format `anthropic/<model-id>`. Look up the latest model IDs at https://docs.anthropic.com/en/docs/about-claude/models.

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

// BaseURL: "https://custom-endpoint/v1", // for non-OpenAI providers

ollamaPlugin := &ollama.Ollama{ 	ServerAddress: "http://localhost:11434",

ServerAddress: "http://localhost:11434",

The `@genkit-ai/a2ui` plugin brings > [A2UI](https://a2ui.org/), a transport-agnostic, JSON-based streaming UI

> [A2UI](https://a2ui.org/), a transport-agnostic, JSON-based streaming UI

To render surfaces in the browser you also need a renderer. A2UI ships renderers for [`@a2ui/lit`](https://www.npmjs.com/package/@a2ui/lit),

for [`@a2ui/lit`](https://www.npmjs.com/package/@a2ui/lit),

[`@a2ui/react`](https://www.npmjs.com/package/@a2ui/react), and

[`@a2ui/angular`](https://www.npmjs.com/package/@a2ui/angular). The examples here

<link rel="stylesheet"     href="https://fonts.googleapis.com/css2?family=Material+Symbols+Outlined:opsz,wght,FILL,GRAD@24,400,0..1,0" />

href="https://fonts.googleapis.com/css2?family=Material+Symbols+Outlined:opsz,wght,FILL,GRAD@24,400,0..1,0" />

const myCatalog: A2uiCatalog = {   id: 'https://my-app.org/catalogs/weather.json',

id: 'https://my-app.org/catalogs/weather.json',

### Web Fetch (`@genkit-ai/fetch`)

const weather = remoteAgent({ url: 'http://localhost:8080/api/weatherAgent' });

prompt: [     { text: "change the person's outfit to a banana costume" },     { media: { url: "https://example.com/photo.jpg" } },

{ text: "change the person's outfit to a banana costume" },     { media: { url: "https://example.com/photo.jpg" } },   ],

{ media: { url: "https://example.com/photo.jpg" } },

8. Add the `genkit-cli` to the project's `devDependencies` (e.g. `npm install -D genkit-cli`, or the pnpm/yarn/bun equivalent) so the CLI version is pinned per-project and works in CI, then add a `genkit:dev` script to `package.json` that r

[Credential-related source omitted]

[Credential-related source omitted]

- Python **3.10+** and **`uv`** ([install](https://docs.astral.sh/uv/getting-started/installation/))

[Credential-related source omitted]

client = remote_agent(     url='http://127.0.0.1:8080/api/weatherAgent',

url='http://127.0.0.1:8080/api/weatherAgent',

client = remote_agent(     url='http://127.0.0.1:8080/api/supportAgent',

url='http://127.0.0.1:8080/api/supportAgent',

[Credential-related source omitted]

[Credential-related source omitted]

If the developer doesn't have one: > Get a free key at https://aistudio.google.com/apikey — click **"Create API key"**, copy it.

> Get a free key at https://aistudio.google.com/apikey — click **"Create API key"**, copy it.

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

Genkit Tools UI: http://localhost:4000

The Dev UI is now running at **http://localhost:4000**

1. Open **http://localhost:4000** in your browser

[Credential-related source omitted]

`genkit start --port 4001 -- uv run src/main.py` - `uv: command not found` —   `curl -LsSf https://astral.sh/uv/install.sh | sh`

- `uv: command not found` —   `curl -LsSf https://astral.sh/uv/install.sh | sh` - Flow not showing in Dev UI — check `genkit start` output for errors

`curl -LsSf https://astral.sh/uv/install.sh | sh`

`curl -LsSf https://astral.sh/uv/install.sh | sh`

[Credential-related source omitted]

image_url = 'https://example.com/cat.jpg'  # or data:image/png;base64,...

[Credential-related source omitted]

const res = await fetch('/api/chat', {

**curl test:**

curl -N -X POST http://localhost:8080/api/chat \

[Credential-related source omitted]

[Credential-related source omitted]

Genkit Developer UI: http://localhost:4000

async with httpx.AsyncClient(base_url='http://127.0.0.1:18524') as client:

[Credential-related source omitted]