Skip to main content

examples.tools.ag_ui

AG-UI example tool implementations.

These callables replicate the exact feature set of the former /api/v1/examples/* AG-UI demo agents, exposed as reusable runtime tools that can be attached to agents through agentspecs.

Each function is referenced from a ToolSpec runtime block via package + method and registered on a pydantic_ai.Agent through register_agent_tools (which wires them with tool_plain).

current_time​

async def current_time(timezone: str = "UTC") -> str

Get the current time in ISO format.

Arguments:

  • timezone - The timezone to use (e.g., 'UTC', 'America/New_York', 'Europe/London').

Returns:

The current time in ISO format string.

get_weather​

async def get_weather(location: str) -> dict[str, str | float]

Get current weather for a location.

This tool fetches real weather data from Open-Meteo API. The frontend can render this data as a weather card.

Arguments:

  • location - City name (e.g., "New York", "London", "Tokyo").

Returns:

Dictionary with weather information:

  • temperature: Current temperature in Celsius
  • feelsLike: Apparent temperature
  • humidity: Relative humidity percentage
  • windSpeed: Wind speed in km/h
  • windGust: Wind gust speed in km/h
  • conditions: Human-readable weather description
  • location: Resolved location name

generate_haiku​

async def generate_haiku(japanese: list[str], english: list[str],
                         gradient: str) -> str

Generate a haiku and display it in the UI.

This tool creates a haiku with Japanese text, English translation, and a beautiful gradient background. The frontend will render this as a card in both the chat and the main display area.

Arguments:

  • japanese - Array of three lines of the haiku in Japanese (5-7-5 syllables).
  • english - Array of three lines of the haiku translated to English.
  • gradient - CSS gradient string for the card background.
  • Example - linear-gradient(135deg, color1 0%, color2 100%)

Returns:

Confirmation message.

Step Objects​

class Step(BaseModel)

A step in a plan.

Plan Objects​

class Plan(BaseModel)

A plan with multiple steps.

create_plan​

async def create_plan(steps: list[str]) -> StateSnapshotEvent

Create a plan with multiple steps.

This initializes the shared state with a new plan.

Arguments:

  • steps - List of step descriptions to create the plan.

Returns:

StateSnapshotEvent containing the initial plan state.

update_plan_step​

async def update_plan_step(
        index: int,
        description: Optional[str] = None,
        status: Optional[StepStatus] = None) -> StateDeltaEvent

Update a specific step in the plan.

Uses JSON Patch (RFC 6902) for efficient incremental updates.

Arguments:

  • index - The index of the step to update (0-based).
  • description - New description for the step (optional).
  • status - New status for the step (optional).

Returns:

StateDeltaEvent containing the JSON Patch operations.

TaskStep Objects​

class TaskStep(BaseModel)

A step in a task plan.

TaskPlan Objects​

class TaskPlan(BaseModel)

A task plan with multiple steps for human review.

generate_task_steps​

async def generate_task_steps(steps: list[str]) -> StateSnapshotEvent

Generate a list of task steps for the user to review and approve.

This creates a task plan that will be displayed to the user. The user can enable/disable steps before confirming execution.

Arguments:

  • steps - List of step descriptions (brief imperative commands).

Returns:

StateSnapshotEvent containing the task plan for user review.

SkillLevel Objects​

class SkillLevel(StrEnum)

The skill level required for the recipe.

SpecialPreferences Objects​

class SpecialPreferences(StrEnum)

Special preferences for the recipe.

CookingTime Objects​

class CookingTime(StrEnum)

The cooking time of the recipe.

Ingredient Objects​

class Ingredient(BaseModel)

An ingredient in a recipe.

Recipe Objects​

class Recipe(BaseModel)

A recipe with all its details.

display_recipe​

async def display_recipe(recipe: Recipe) -> StateSnapshotEvent

Display the recipe to the user.

This tool updates the shared state with the new recipe, which is then reflected in the UI.

Arguments:

  • recipe - The complete recipe to display.

Returns:

StateSnapshotEvent containing the recipe snapshot.