@cuylabs/agent-foundry-hosting

@cuylabs/agent-foundry-hosting

Foundry Hosting integration for @cuylabs/agent-core.

This package mirrors Python agent-framework-foundry-hosting. It is the higher-level adapter layer that wires an agent framework runtime into the lower-level Agent Server protocol packages:

@cuylabs/agent-foundry-agentserver-invocations  -> azure-ai-agentserver-invocations
@cuylabs/agent-foundry-agentserver-responses    -> azure-ai-agentserver-responses
@cuylabs/agent-foundry-hosting                  -> agent-framework-foundry-hosting

Invocations

import { runInvocationsServer } from "@cuylabs/agent-foundry-agentserver-invocations";
import { createInvocationsHandlerForAgent } from "@cuylabs/agent-foundry-hosting/invocations";
import {
  InProcessAgentServer,
} from "@cuylabs/agent-server";
import { createAgentCoreServerAdapter } from "@cuylabs/agent-server/agent-core";

const server = new InProcessAgentServer(createAgentCoreServerAdapter(myAgent));

await runInvocationsServer({
  handler: createInvocationsHandlerForAgent(server),
  port: 8088,
});

Responses

import { runResponsesServer } from "@cuylabs/agent-foundry-agentserver-responses";
import { createResponsesHandlerForAgent } from "@cuylabs/agent-foundry-hosting/responses";
import {
  InProcessAgentServer,
} from "@cuylabs/agent-server";
import { createAgentCoreServerAdapter } from "@cuylabs/agent-server/agent-core";

const server = new InProcessAgentServer(createAgentCoreServerAdapter(myAgent));

await runResponsesServer({
  handler: createResponsesHandlerForAgent(server),
  port: 8088,
});

The Responses bridge starts an agent-server turn for each create-response request, maps agent-core text/tool/error/completion events into Responses stream events, and leaves HTTP, SSE framing, IDs, storage, and platform headers to @cuylabs/agent-foundry-agentserver-responses.