PydanticAI Platform Client

A lightweight Python client for the PydanticAI Multi-Agent Platform API.

Installation

pip install pydanticai-platform-client

Or install from source:

pip install -e sdk/

Quick Start

from pydanticai_platform_client import PlatformClient

async def main():
    async with PlatformClient(
        base_url="https://your-platform.fly.dev",
        api_key="pk_your_tenant_key",
    ) as client:
        # Chat with the default agent
        response = await client.chat("What's the weather like?")
        print(response.text)

        # Use a specific agent
        response = await client.chat(
            "Analyze this data: [1, 2, 3, 4, 5]",
            agent="analyst",
        )
        print(response.text)

        # Continue a conversation
        response = await client.chat(
            "Tell me more",
            conversation_id=response.conversation_id,
        )

Streaming Responses

async with PlatformClient(base_url, api_key) as client:
    async for chunk in client.chat_stream("Write a poem about Python"):
        print(chunk, end="", flush=True)
    print()  # newline at the end

Available Methods

Error Handling

from pydanticai_platform_client import (
    PlatformClient,
    AuthenticationError,
    RateLimitError,
    TokenLimitError,
    AgentNotFoundError,
)

async with PlatformClient(base_url, api_key) as client:
    try:
        response = await client.chat("Hello")
    except AuthenticationError:
        print("Invalid API key")
    except RateLimitError as e:
        print(f"Rate limited, retry after {e.retry_after}s")
    except TokenLimitError as e:
        print(f"Token limit exceeded: {e.tokens_used}/{e.limit}")
    except AgentNotFoundError as e:
        print(f"Agent not found: {e.agent_name}")

Response Models

ChatResponse

@dataclass
class ChatResponse:
    text: str              # The agent's response
    conversation_id: str   # ID for continuing the conversation
    agent: str             # Agent that handled the request
    model: str             # Model used

UsageStats

@dataclass
class UsageStats:
    tenant_id: str
    period_days: int
    total_requests: int
    total_tokens: int
    estimated_cost_usd: float
    by_model: dict[str, int]
    by_agent: dict[str, int]

License

MIT