Demo

python/packages/autogen-studio/Dockerfile

@@ -1,17 +1,33 @@
 FROM python:3.10-slim
-WORKDIR /code
 
-RUN pip install -U gunicorn autogenstudio
+# 1. Install build-tools if you have any compiled deps
+RUN apt-get update \
+ && apt-get install -y --no-install-recommends build-essential \
+ && rm -rf /var/lib/apt/lists/*
 
-# Create a non-root user
+# 2. Create and switch to a non-root user
 RUN useradd -m -u 1000 user
 USER user
 ENV HOME=/home/user \
     PATH=/home/user/.local/bin:$PATH \
     AUTOGENSTUDIO_APPDIR=/home/user/app
 
+# 3. Copy your local package metadata first (for caching)
 WORKDIR $HOME/app
+COPY --chown=user . .
 
-COPY --chown=user . $HOME/app
+# 4. Install Gunicorn (in case your CLI/UI entrypoint uses it)
+RUN pip install --upgrade pip \
+ && pip install --user gunicorn uvicorn
 
-CMD gunicorn -w $((2 * $(getconf _NPROCESSORS_ONLN) + 1)) --timeout 12600 -k uvicorn.workers.UvicornWorker autogenstudio.web.app:app --bind "0.0.0.0:8081"
+# 5. Install your package in editable mode
+RUN pip install --user -e .
+
+# 6. Expose port and set the default CMD
+EXPOSE 8081
+CMD ["python", "-m", "gunicorn", \
+     "-w", "4", \
+     "--timeout", "12600", \
+     "-k", "uvicorn.workers.UvicornWorker", \
+     "autogenstudio.web.app:app", \
+     "--bind", "0.0.0.0:8081"]

python/packages/autogen-studio/autogenstudio/gallery/builder.py

@@ -142,7 +142,7 @@ def create_default_gallery() -> GalleryConfig:
             os.environ[key] = "test"
 
     # url = "https://raw.githubusercontent.com/microsoft/autogen/refs/heads/main/python/packages/autogen-studio/autogenstudio/gallery/default.json"
-    builder = GalleryBuilder(id="gallery_default", name="Default Component Gallery")
+    builder = GalleryBuilder(id="collection_default", name="Default Component Collection")
 
     # Set metadata
     builder.set_metadata(

python/packages/autogen-studio/autogenstudio/utils/utils.py

@@ -1,13 +1,29 @@
 import base64
-from typing import Sequence
-
+import io
+import os
+from typing import Sequence, Optional
+from ..datamodel import Run
 from autogen_agentchat.messages import ChatMessage, MultiModalMessage, TextMessage
 from autogen_core import Image
 from autogen_core.models import UserMessage
 from loguru import logger
 
+from azure.core.credentials import AzureKeyCredential
+from azure.ai.formrecognizer import DocumentAnalysisClient
+
+def _get_run(self, run_id: int) -> Optional[Run]:
+    """Get run from database
+
+    Args:
+        run_id: id of the run to retrieve
+
+    Returns:
+        Optional[Run]: Run object if found, None otherwise
+    """
+    response = self.db_manager.get(Run, filters={"id": run_id}, return_json=False)
+    return response.data[0] if response.status and response.data else None
 
-def construct_task(query: str, files: list[dict] | None = None) -> Sequence[ChatMessage]:
+def construct_task(query: str, files: list[dict] | None = None, env_vars: Optional[list[dict]] = None) -> Sequence[ChatMessage]:
     """
     Construct a task from a query string and list of files.
     Returns a list of ChatMessage objects suitable for processing by the agent system.
@@ -28,9 +44,25 @@ def construct_task(query: str, files: list[dict] | None = None) -> Sequence[Chat
     if query:
         messages.append(TextMessage(source="user", content=query))
 
+    for env_var in env_vars or []:
+        if env_var.name == "DOCUMENT_INTEL_ENDPOINT":
+            DOCUMENT_INTEL_ENDPOINT = env_var.value
+        elif env_var.name == "DOCUMENT_INTEL_KEY":
+            DOCUMENT_INTEL_KEY = env_var.value
+
     # Process each file based on its type
     for file in files:
         try:
+            ftype = file.get("type", "")
+            fname = file.get("name", "unknown")
+            content_b64 = file["content"]
+
+            doc_intel_client = DocumentAnalysisClient(
+                endpoint=DOCUMENT_INTEL_ENDPOINT,
+                credential=AzureKeyCredential(DOCUMENT_INTEL_KEY),
+            )
+
+
             if file.get("type", "").startswith("image/"):
                 # Handle image file using from_base64 method
                 # The content is already base64 encoded according to the convertFilesToBase64 function
@@ -48,6 +80,34 @@ def construct_task(query: str, files: list[dict] | None = None) -> Sequence[Chat
                         source="user", content=text_content, metadata={"filename": file.get("name", "unknown.txt")}
                     )
                 )
+            elif ftype == "application/pdf":
+                # 1) Decode the PDF bytes
+                pdf_bytes = base64.b64decode(content_b64)
+                stream = io.BytesIO(pdf_bytes)
+
+                # 2) Invoke Azure’s “prebuilt-document” model to get structured text
+                poller = doc_intel_client.begin_analyze_document(
+                    "prebuilt-document",
+                    document=stream
+                )
+                result = poller.result()
+
+                # 3) Concatenate everything into one big string (you could also
+                #    chunk it by page or section, depending on your needs)
+                full_text = []
+                for page in result.pages:
+                    for line in page.lines:
+                        full_text.append(line.content)
+                document_text = "\n".join(full_text)
+
+                messages.append(
+                    TextMessage(
+                        source="user",
+                        content=document_text,
+                        metadata={"filename": fname, "filetype": ftype},
+                    )
+                )
+
             else:
                 # Log unsupported file types but still try to process based on best guess
                 logger.warning(f"Potentially unsupported file type: {file.get('type')} for file {file.get('name')}")
@@ -69,3 +129,37 @@ def construct_task(query: str, files: list[dict] | None = None) -> Sequence[Chat
             # Continue processing other files even if one fails
 
     return messages
+
+
+def update_team_config(team_config: dict, tasks: Sequence[ChatMessage]) -> dict:
+    """
+    Build and inject the selector_prompt into team_config based on incoming tasks.
+    Returns the updated team_config.
+    """
+    cfg      = team_config["config"]
+    template = cfg.get("selector_prompt", "You are a helpful assistant.")
+
+    # extract first matching TextMessage for context and form
+    context = next(
+        (t.content for t in tasks
+         if isinstance(t, TextMessage) and "context" in t.metadata.get("filename", "").lower()),
+        ""
+    )
+    form = next(
+        (t.content for t in tasks
+         if isinstance(t, TextMessage) and "form" in t.metadata.get("filename", "").lower()),
+        ""
+    )
+
+    # append any multimodal filenames
+    multimodal_files = [
+        t.metadata.get("filename", "unknown.file")
+        for t in tasks
+        if isinstance(t, MultiModalMessage)
+    ]
+    if multimodal_files:
+        template += "\nIncluded files: " + ", ".join(multimodal_files)
+
+    # format and store back into the config
+    cfg["selector_prompt"] = template.format(context=context, form=form)
+    return team_config

python/packages/autogen-studio/autogenstudio/web/routes/ws.py

@@ -2,13 +2,14 @@
 import asyncio
 import json
 from datetime import datetime
+from typing import Optional
 
 from fastapi import APIRouter, Depends, WebSocket, WebSocketDisconnect
 from fastapi.websockets import WebSocketState
 from loguru import logger
 
-from ...datamodel import Run, RunStatus
-from ...utils.utils import construct_task
+from ...datamodel import Run, RunStatus, Settings, SettingsConfig
+from ...utils.utils import construct_task, update_team_config
 from ..auth.dependencies import get_ws_auth_manager
 from ..auth.wsauth import WebSocketAuthHandler
 from ..deps import get_db, get_websocket_manager
@@ -35,6 +36,18 @@ async def run_websocket(
             return
 
         run = run_response.data[0]
+        async def _get_settings(user_id: str) -> Optional[Settings]:
+            """Get user settings from database
+            Args:
+                user_id: User ID to retrieve settings for
+            Returns:
+                Optional[dict]: User settings if found, None otherwise
+            """
+            response = db.get(filters={"user_id": user_id}, model_class=Settings, return_json=False)
+            return response.data[0] if response.status and response.data else None
+
+        user_settings = await _get_settings(run.user_id)
+        env_vars = SettingsConfig(**user_settings.config).environment if user_settings else None  # type: ignore
 
         if run.status not in [RunStatus.CREATED, RunStatus.ACTIVE]:
             await websocket.close(code=4003, reason="Run not in valid state")
@@ -84,9 +97,10 @@ async def run_websocket(
                 if message.get("type") == "start":
                     # Handle start message
                     logger.info(f"Received start request for run {run_id}")
-                    task = construct_task(query=message.get("task"), files=message.get("files"))
+                    task = construct_task(query=message.get("task"), files=message.get("files"), env_vars=env_vars)
 
                     team_config = message.get("team_config")
+                    team_config = update_team_config(team_config, task)
                     if task and team_config:
                         # Start the stream in a separate task
                         asyncio.create_task(ws_manager.start_stream(run_id, task, team_config))

python/packages/autogen-studio/frontend/gatsby-config.ts

@@ -13,12 +13,13 @@ require("dotenv").config({
   path: envFile,
 });
 
+
 const config: GatsbyConfig = {
   pathPrefix: process.env.PREFIX_PATH_VALUE || "",
   siteMetadata: {
-    title: `AutoGen Studio`,
-    description: `Build Multi-Agent Apps`,
-    siteUrl: `http://tbd.place`,
+    title: `AI Planet Studio`,
+    description: `Build Multi-Agent Worflows & Apps`,
+    siteUrl: `http://www.aiplanet.com`,
   },
   // More easily incorporate content into your pages through automatic TypeScript type generation and better GraphQL IntelliSense.
   // If you use VSCode you can also use the GraphQL plugin

python/packages/autogen-studio/frontend/src/components/footer.tsx

@@ -19,16 +19,16 @@ const Footer = () => {
   return (
     <div className=" text-primary p-3  border-t border-secondary flex ">
       <div className="text-xs flex-1">
-        Maintained by the AutoGen{" "}
-        <a
+        Maintained by the AI Planet Team.{" "}
+        {/* <a
           target={"_blank"}
           rel={"noopener noreferrer"}
           className="underlipne inline-block border-accent border-b hover:text-accent"
           href="https://microsoft.github.io/autogen/"
         >
           {" "}
           Team.
-        </a>
+        </a> */}
       </div>
       {version && (
         <div className="text-xs ml-2 text-secondary"> v{version}</div>