Add server implementation

Dockerfile

@@ -1,13 +1,30 @@
-FROM quay.ocp.banorte.com/golden/python-312:latest
+FROM quay.ocp.banorte.com/golden/python-312:latest AS builder
 
-COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/
+COPY --from=ghcr.io/astral-sh/uv:0.7.12 /uv /uvx /bin/
+
+ENV UV_COMPILE_BYTECODE=1 \
+    UV_NO_CACHE=1 \
+    UV_NO_DEV=1 \
+    UV_LINK_MODE=copy
 
 WORKDIR /app
 
-COPY . .
+# Install dependencies first (cached layer as long as lockfile doesn't change)
+COPY pyproject.toml uv.lock ./
+RUN uv sync --locked --no-install-project --no-editable
 
-RUN uv sync
+# Copy the rest of the project and install it
+COPY . .
+RUN uv sync --locked --no-editable
+
+# --- Final stage: no uv, no build artifacts ---
+FROM quay.ocp.banorte.com/golden/python-312:latest
+
+WORKDIR /app
+
+COPY --from=builder /app/.venv /app/.venv
+COPY config.yaml ./
 
 ENV PATH="/app/.venv/bin:$PATH"
 
-CMD ["uv", "run", "uvicorn", "rag_eval.server:app", "--host", "0.0.0.0"]
+CMD ["uvicorn", "va_agent.server:app", "--host", "0.0.0.0"]

src/va_agent/server.py

@@ -1,10 +1,142 @@
-"""FastAPI server exposing the RAG agent endpoint.
+"""FastAPI server exposing the RAG agent endpoint."""
 
-NOTE: This file is a stub. The rag_eval module was removed in the
-lean MCP implementation. This file is kept for reference but is not
-functional.
-"""
+from __future__ import annotations
 
-from fastapi import FastAPI
+import logging
+import uuid
+from typing import Any
 
-app = FastAPI(title="RAG Agent")
+from fastapi import FastAPI, HTTPException
+from google.genai.types import Content, Part
+from pydantic import BaseModel, Field
+
+from va_agent.agent import runner
+
+logger = logging.getLogger(__name__)
+
+app = FastAPI(title="Vaia Agent")
+
+
+# ---------------------------------------------------------------------------
+# Request / Response models
+# ---------------------------------------------------------------------------
+
+
+class NotificationPayload(BaseModel):
+    """Notification context sent alongside a user query."""
+
+    text: str | None = None
+    parameters: dict[str, Any] = Field(default_factory=dict)
+
+
+class QueryRequest(BaseModel):
+    """Incoming query request from the integration layer."""
+
+    phone_number: str
+    text: str
+    type: str = "conversation"
+    notification: NotificationPayload | None = None
+    language_code: str = "es"
+
+
+class QueryResponse(BaseModel):
+    """Response returned to the integration layer."""
+
+    response_id: str
+    response_text: str
+    parameters: dict[str, Any] = Field(default_factory=dict)
+    confidence: float | None = None
+
+
+class ErrorResponse(BaseModel):
+    """Standard error body."""
+
+    error: str
+    message: str
+    status: int
+
+
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+
+
+def _build_user_message(request: QueryRequest) -> str:
+    """Compose the text sent to the agent, including notification context."""
+    if request.type == "notification" and request.notification:
+        parts = [request.text]
+        if request.notification.text:
+            parts.append(
+                f"\n[Notificación recibida]: {request.notification.text}"
+            )
+        if request.notification.parameters:
+            formatted = ", ".join(
+                f"{k}: {v}"
+                for k, v in request.notification.parameters.items()
+            )
+            parts.append(f"[Parámetros de notificación]: {formatted}")
+        return "\n".join(parts)
+    return request.text
+
+
+# ---------------------------------------------------------------------------
+# Endpoints
+# ---------------------------------------------------------------------------
+
+
+@app.post(
+    "/api/v1/query",
+    response_model=QueryResponse,
+    responses={
+        400: {"model": ErrorResponse},
+        500: {"model": ErrorResponse},
+        503: {"model": ErrorResponse},
+    },
+)
+async def query(request: QueryRequest) -> QueryResponse:
+    """Process a user message and return a generated response."""
+    user_message = _build_user_message(request)
+    session_id = request.phone_number
+    user_id = request.phone_number
+
+    new_message = Content(
+        role="user",
+        parts=[Part(text=user_message)],
+    )
+
+    try:
+        response_text = ""
+        async for event in runner.run_async(
+            user_id=user_id,
+            session_id=session_id,
+            new_message=new_message,
+        ):
+            if event.content and event.content.parts:
+                for part in event.content.parts:
+                    if part.text and event.author != "user":
+                        response_text += part.text
+    except ValueError as exc:
+        logger.exception("Bad request while running agent")
+        raise HTTPException(
+            status_code=400,
+            detail=ErrorResponse(
+                error="Bad Request",
+                message=str(exc),
+                status=400,
+            ).model_dump(),
+        ) from exc
+    except Exception as exc:
+        logger.exception("Internal error while running agent")
+        raise HTTPException(
+            status_code=500,
+            detail=ErrorResponse(
+                error="Internal Server Error",
+                message="Failed to generate response",
+                status=500,
+            ).model_dump(),
+        ) from exc
+
+    return QueryResponse(
+        response_id=f"rag-resp-{uuid.uuid4()}",
+        response_text=response_text,
+    )