Feat: Generative UI prototype [DO NOT MERGE]

agents/generative-interface/Dockerfile

@@ -0,0 +1,11 @@
+FROM python:3.13-alpine3.22
+ARG RELEASE_VERSION="main"
+COPY ./agents/generative-interface/ /app/agents/generative-interface
+COPY ./apps/agentstack-sdk-py/ /app/apps/agentstack-sdk-py/
+WORKDIR /app/agents/generative-interface
+RUN --mount=type=cache,target=/tmp/.cache/uv \
+    --mount=type=bind,from=ghcr.io/astral-sh/uv:0.9.5,source=/uv,target=/bin/uv \
+    UV_COMPILE_BYTECODE=1 HOME=/tmp uv sync
+ENV PRODUCTION_MODE=True \
+    RELEASE_VERSION=${RELEASE_VERSION}
+CMD ["/app/agents/generative-interface/.venv/bin/server"]

agents/generative-interface/pyproject.toml

@@ -0,0 +1,36 @@
+[project]
+name = "generative-interface"
+version = "0.1.0"
+description = "Agent demonstrating generative interface extension."
+authors = [
+    { name = "IBM Corp." },
+]
+requires-python = ">=3.13,<3.14"
+dependencies = [
+    "agentstack-sdk",
+    "openinference-instrumentation-beeai>=0.1.14",
+    "pydantic-settings>=2.9.0",
+    "uvicorn>=0.35.0",
+]
+
+[tool.ruff]
+line-length = 120
+
+[tool.uv.sources]
+agentstack-sdk = { path = "../../apps/agentstack-sdk-py", editable = true }
+
+[project.scripts]
+server = "generative_interface.agent:serve"
+
+[build-system]
+requires = ["uv_build>=0.9.0,<0.10.0"]
+build-backend = "uv_build"
+
+[dependency-groups]
+dev = [
+    "watchfiles>=1.1.0",
+]
+
+[tool.pyright]
+venvPath = "."
+venv = ".venv"

agents/generative-interface/src/generative_interface/__init__.py

@@ -0,0 +1,6 @@
+# Copyright 2025 © BeeAI a Series of LF Projects, LLC
+# SPDX-License-Identifier: Apache-2.0
+
+from .agent import serve
+
+__all__ = ["serve"]

agents/generative-interface/src/generative_interface/agent.py

@@ -0,0 +1,128 @@
+# Copyright 2025 © BeeAI a Series of LF Projects, LLC
+# SPDX-License-Identifier: Apache-2.0
+
+import json
+import os
+from typing import Annotated
+
+from a2a.types import Message, TextPart
+from agentstack_sdk.a2a.extensions import (
+    AgentDetail,
+    AgentDetailContributor,
+    LLMServiceExtensionServer,
+    LLMServiceExtensionSpec,
+)
+from agentstack_sdk.a2a.extensions.services.generative_interface import (
+    GenerativeInterfaceExtensionServer,
+    GenerativeInterfaceExtensionSpec,
+    GenerativeInterfaceSpec,
+)
+from agentstack_sdk.server import Server
+from agentstack_sdk.server.context import RunContext
+from agentstack_sdk.server.middleware.platform_auth_backend import PlatformAuthBackend
+from openai import AsyncOpenAI
+from openai.types.chat import ChatCompletionMessageParam
+
+server = Server()
+
+AGENT_GOAL = """You are a financial advisor assistant. Your first task is to ask the user about their gender (male/female/fish) before providing any financial advice. Use the UI components to create an interactive experience."""
+
+
+def _get_text(message: Message) -> str:
+    return "\n\n".join(part.root.text for part in message.parts or [] if isinstance(part.root, TextPart))
+
+
+@server.agent(
+    name="Generative Interface Agent",
+    documentation_url=f"https://github.com/i-am-bee/agentstack/blob/{os.getenv('RELEASE_VERSION', 'main')}/agents/generative-interface",
+    version="1.0.0",
+    default_input_modes=["text", "text/plain"],
+    default_output_modes=["text", "text/plain"],
+    description="Financial advisor with dynamic UI generation",
+    detail=AgentDetail(
+        interaction_mode="multi-turn",
+        author=AgentDetailContributor(name="IBM"),
+    ),
+)
+async def agent(
+    message: Message,
+    context: RunContext,
+    ui: Annotated[GenerativeInterfaceExtensionServer, GenerativeInterfaceExtensionSpec.demand()],
+    llm: Annotated[LLMServiceExtensionServer, LLMServiceExtensionSpec.single_demand()],
+):
+    await context.store(message)
+
+    (llm_config,) = llm.data.llm_fulfillments.values()
+    client = AsyncOpenAI(
+        api_key=llm_config.api_key,
+        base_url=llm_config.api_base,
+    )
+
+    system_prompt = f"""{ui.catalog_prompt}
+
+{AGENT_GOAL}
+"""
+
+    history = context.load_history()
+    llm_messages: list[ChatCompletionMessageParam] = [{"role": "system", "content": system_prompt}]
+
+    async for item in history:
+        if isinstance(item, Message):
+            if content := _get_text(item):
+                role = "assistant" if item.role == "agent" else "user"
+                llm_messages.append({"role": role, "content": content})
+
+    response = await client.chat.completions.create(
+        model=llm_config.api_model,
+        messages=llm_messages,
+    )
+
+    assistant_content = response.choices[0].message.content or ""
+
+    ui_spec = parse_spec_stream(assistant_content)
+
+    if ui_spec:
+        ui_response = await ui.request_ui(spec=ui_spec)
+        if ui_response:
+            yield f"You selected: {ui_response.component_id}"
+
+
+def parse_spec_stream(content: str) -> GenerativeInterfaceSpec | None:
+    spec: dict = {"root": "", "elements": {}}
+
+    for line in content.strip().split("\n"):
+        line = line.strip()
+        if not line:
+            continue
+        try:
+            patch = json.loads(line)
+            if patch.get("op") == "set":
+                path = patch.get("path", "")
+                value = patch.get("value")
+                if path == "/root":
+                    spec["root"] = value
+                elif path.startswith("/elements/"):
+                    key = path[len("/elements/"):]
+                    spec["elements"][key] = value
+        except json.JSONDecodeError:
+            continue
+
+    if spec["root"] and spec["elements"]:
+        return GenerativeInterfaceSpec.model_validate(spec)
+    return None
+
+
+def serve():
+    try:
+        server.run(
+            host=os.getenv("HOST", "127.0.0.1"),
+            port=int(os.getenv("PORT", 10001)),
+            configure_telemetry=True,
+            auth_backend=PlatformAuthBackend(),
+        )
+    except KeyboardInterrupt:
+        pass
+
+
+if __name__ == "__main__":
+    serve()