Rewrite Modal app with inline Gradio for ASGI compatibility

Avoid file import issues and SSE streaming problems by defining
the Gradio app inline in the Modal function.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

modal_app.py

@@ -31,18 +31,120 @@ image = (
         modal.Secret.from_name("elevenlabs-api-key"),
     ],
     timeout=600,
-    container_idle_timeout=300,
+    scaledown_window=300,
 )
-@modal.web_server(port=7860, startup_timeout=120)
+@modal.asgi_app()
 def serve():
-    """Serve the Gradio app via web_server."""
-    import subprocess
+    """Serve the Gradio app as ASGI."""
+    import sys
     import os
+    sys.path.insert(0, "/app")
 
     # Disable MCP server on Modal (causes issues)
     os.environ["ENABLE_MCP_SERVER"] = "false"
-    os.chdir("/app")
-    subprocess.Popen(["python", "app.py"])
+
+    import gradio as gr
+    from src.personas import PERSONAS, get_persona_names, get_persona
+    from src.agent import run_agent
+    from src.tts import generate_speech
+    import tempfile
+
+    def format_sources(sources):
+        if not sources:
+            return "*No external sources used*"
+        md = ""
+        for i, src in enumerate(sources, 1):
+            if src.get("url"):
+                md += f"{i}. [{src['title']}]({src['url']})\n"
+            else:
+                md += f"{i}. {src['title']} ({src.get('source', 'General')})\n"
+        return md
+
+    def format_mcp_tools(tools):
+        if not tools:
+            return "*No tools used*"
+        md = "**Agent Tool Calls:**\n\n"
+        for tool in tools:
+            md += f"| {tool['icon']} | `{tool['name']}` | {tool['desc']} |\n"
+        return md
+
+    def explain_topic(topic, persona_name, audience=""):
+        if not topic.strip():
+            return "Please enter a topic!", "", "", ""
+        if not persona_name:
+            persona_name = "5-Year-Old"
+
+        steps_log = []
+        explanation = ""
+        sources = []
+        mcp_tools = []
+
+        for update in run_agent(topic, persona_name, audience):
+            if update["type"] == "step":
+                steps_log.append(f"**{update['title']}**\n{update['content']}")
+                if update["step"] == "research_done" and "sources" in update:
+                    sources = update["sources"]
+            elif update["type"] == "result":
+                explanation = update["explanation"]
+                sources = update.get("sources", sources)
+                mcp_tools = update.get("mcp_tools", [])
+
+        return explanation, format_sources(sources), "\n\n---\n\n".join(steps_log), format_mcp_tools(mcp_tools)
+
+    def generate_audio(explanation, persona_name):
+        if not explanation or not explanation.strip():
+            return None
+        if not persona_name:
+            persona_name = "5-Year-Old"
+        persona = get_persona(persona_name)
+        voice_id = persona["voice_id"]
+        voice_settings = persona.get("voice_settings")
+        audio_bytes = generate_speech(explanation, voice_id, voice_settings)
+        with tempfile.NamedTemporaryFile(suffix=".mp3", delete=False) as f:
+            f.write(audio_bytes)
+            return f.name
+
+    with gr.Blocks(title="Explainor - AI Persona Explanations") as demo:
+        gr.Markdown("# Explainor\n### Learn anything through the voice of your favorite characters!")
+
+        with gr.Row():
+            topic_input = gr.Textbox(label="What do you want to learn about?", placeholder="e.g., Quantum Computing")
+            persona_choices = [f"{PERSONAS[name]['emoji']} {name}" for name in get_persona_names()]
+            persona_dropdown = gr.Dropdown(choices=persona_choices, value=persona_choices[0], label="Choose your explainer")
+
+        audience_dropdown = gr.Dropdown(
+            choices=["Just me", "My confused grandmother", "A skeptical robot", "An alien visiting Earth"],
+            value="Just me",
+            label="Who's listening?"
+        )
+
+        explain_btn = gr.Button("Explain it to me!", variant="primary")
+
+        explanation_output = gr.Textbox(label="Explanation", lines=6)
+        read_aloud_btn = gr.Button("Read Aloud", variant="secondary")
+        audio_output = gr.Audio(label="Listen", type="filepath", autoplay=True)
+
+        with gr.Accordion("Agent Tool Calls", open=False):
+            mcp_output = gr.Markdown("")
+        with gr.Accordion("Sources", open=False):
+            sources_output = gr.Markdown("")
+        with gr.Accordion("Execution Trace", open=False):
+            steps_output = gr.Markdown("")
+
+        def process_explain(topic, persona_with_emoji, audience):
+            persona_name = persona_with_emoji.split(" ", 1)[1] if " " in persona_with_emoji else persona_with_emoji
+            aud = "" if "Just me" in audience else audience
+            return explain_topic(topic, persona_name, aud)
+
+        def process_audio(explanation, persona_with_emoji):
+            persona_name = persona_with_emoji.split(" ", 1)[1] if " " in persona_with_emoji else persona_with_emoji
+            return generate_audio(explanation, persona_name)
+
+        explain_btn.click(fn=process_explain, inputs=[topic_input, persona_dropdown, audience_dropdown],
+                          outputs=[explanation_output, sources_output, steps_output, mcp_output])
+        read_aloud_btn.click(fn=process_audio, inputs=[explanation_output, persona_dropdown], outputs=[audio_output])
+
+    return demo.queue()
 
 
 # For local testing