Skip to main content

routes.vercel_ai

Vercel AI SDK routes for agent-runtimes server.

TerminateRequest Objects​

class TerminateRequest(BaseModel)

Request to terminate a running Vercel AI request.

TerminateResponse Objects​

class TerminateResponse(BaseModel)

Response from terminate request.

register_request​

def register_request(request_id: str) -> asyncio.Event

Register a running request and return its cancellation event.

Arguments:

  • request_id - The unique request identifier.

Returns:

An asyncio.Event that can be set to signal cancellation.

unregister_request​

def unregister_request(request_id: str) -> None

Unregister a request when it completes.

Arguments:

  • request_id - The request identifier to remove.

cancel_request​

def cancel_request(request_id: str) -> bool

Cancel a running request.

Arguments:

  • request_id - The request identifier to cancel.

Returns:

True if the request was found and cancelled, False otherwise.

cancel_all_requests​

def cancel_all_requests() -> int

Cancel all running requests.

Returns:

Number of requests cancelled.

register_vercel_agent​

def register_vercel_agent(agent_id: str, adapter: VercelAITransport) -> None

Register a Vercel AI adapter.

Arguments:

  • agent_id - Unique identifier for the agent.
  • adapter - The VercelAITransport instance.

unregister_vercel_agent​

def unregister_vercel_agent(agent_id: str) -> None

Unregister a Vercel AI adapter.

Arguments:

  • agent_id - The agent identifier.

get_vercel_adapter​

def get_vercel_adapter(agent_id: str) -> VercelAITransport | None

Get a Vercel AI adapter by ID.

Arguments:

  • agent_id - The agent identifier.

Returns:

The VercelAITransport if found, None otherwise.

list_agents​

@router.get("/agents")
async def list_agents() -> dict[str, Any]

List available Vercel AI agents.

Returns:

Dictionary with list of agent IDs.

terminate_agent​

@router.post("/terminate", response_model=TerminateResponse)
async def terminate_agent(request: TerminateRequest) -> TerminateResponse

Terminate a running Vercel AI request or all requests.

This endpoint allows clients to stop running agent executions. If request_id is provided, only that request is cancelled. If request_id is None, all running requests are cancelled.

Arguments:

  • request - Terminate request with optional request_id.

Returns:

Result of the termination request.

list_requests​

@router.get("/requests")
async def list_requests() -> dict[str, Any]

List all running Vercel AI requests.

Returns:

Dictionary with list of running request IDs.

chat​

@router.post("/{agent_id:path}")
async def chat(request: Request, agent_id: str) -> Response

Handle Vercel AI SDK chat requests.

This endpoint implements the Vercel AI SDK streaming protocol, providing:

  • Streaming chat responses
  • Tool call support
  • Token usage tracking
  • Standard message format

The model can be specified in the request body to override the agent's default.

Arguments:

  • request - The FastAPI/Starlette request.
  • agent_id - The agent to use (defaults to "demo-agent").

Returns:

Streaming response compatible with Vercel AI SDK.

Example:

```javascript
// Client-side with Vercel AI SDK
import { useChat } from 'ai/react';

const { messages, input, handleInputChange, handleSubmit } = useChat({
  api: '/api/v1/vercel-ai/chat',
});
```