azure-ai-agentserver-invocations

What it is and what it does

This package extends the azure-ai-agentserver-core framework with a complete invocation protocol implementation for Azure AI Hosted Agent containers. It exposes HTTP endpoints (POST /invocations, GET /invocations/{id}, POST /invocations/{id}/cancel) and a WebSocket endpoint (/invocations_ws) for full-duplex streaming, along with OpenAPI and AsyncAPI documentation endpoints. Handlers are registered via decorators (@app.invoke_handler, @app.get_invocation_handler, @app.cancel_invocation_handler, @app.ws_handler) and receive request context including invocation ID, session ID, user ID, and call ID through request.state.

The package manages request/response headers automatically (x-agent-invocation-id, x-agent-session-id) and integrates with distributed tracing. It supports long-running operations with polling, streaming responses, and multi-user sessions on container protocol 2.0.0 by forwarding per-request call IDs to downstream services. Built on aiohttp, it runs as an async application with minimal boilerplate.

Use it for:

• Build an agent that processes POST /invocations requests and returns JSON responses.
• Implement long-running agent operations with polling via GET /invocations/{id} and cancellation.
• Stream agent responses as Server-Sent Events or full-duplex WebSocket messages to clients.
• Handle multi-user sessions by forwarding x-agent-foundry-call-id headers to toolbox services.
• Expose OpenAPI and AsyncAPI documentation for agent discovery and client code generation.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Provides HTTP and WebSocket invocation protocol endpoints for Azure AI Hosted Agent containers, plugging into the azure-ai-agentserver-core framework to handle agent execution requests and streaming responses.

Yes. The package is actively maintained (release 4 days old), has zero known vulnerabilities, uses a permissive MIT license, and requires only two runtime dependencies with low install friction. It is marked Production/Stable and supports Python 3.10, 3.11, 3.12, 3.13, and 3.14. Install it if you are building an Azure AI Hosted Agent container and need standard invocation protocol endpoints.

Install

azure-ai-agentserver-invocations on PyPI

pip install azure-ai-agentserver-invocations

uv add azure-ai-agentserver-invocations

poetry add azure-ai-agentserver-invocations

Installing azure-ai-agentserver-invocations

Before you install

Low install friction with two runtime dependencies (azure-ai-agentserver-core and aiohttp). Actively maintained with a release 4 days old and recent commits; marked Production/Stable.

License in practice

MIT license permits commercial and private use with minimal restrictions; suitable for most projects.

Quickstart

pip install azure-ai-agentserver-invocations

from azure.ai.agentserver.invocations import InvocationAgentServerHost

app = InvocationAgentServerHost()

@app.invoke_handler
async def handle(request):
    data = await request.json()
    return {"result": data}

app.run()

Requires Python 3.10 or later.

Verify before relying

• Whether the package is suitable for production workloads beyond the stated Azure AI Hosted Agent container use case.
• Performance characteristics under high concurrency or large message volumes on WebSocket connections.
• Specific behavior and guarantees of the distributed tracing integration with GenAI semantic conventions.

Package facts

Evidence: azure_ai_agentserver_invocations-1.0.0-py3-none-any.whl

Keywords: azure, azure sdk, agent, agentserver, invocations

Tags

More Application Frameworks packages

Further reading