GitHub user FrancescoCastaldi added a comment to the discussion: ai chat bot
Integrating an AI chatbot into Apache Superset to assist users with querying
and creating visualizations can be accomplished cleanly using one of two proven
architectures:
### Architecture 1: Portal / Embedded Wrapper (Recommended for Production)
Rather than modifying the Superset frontend source code, host Superset inside
an internal web portal using the `@superset-ui/embedded-sdk`.
1. The host portal embeds the Superset dashboard via iframe using guest tokens.
2. A floating chat drawer or sidebar sits on the host portal next to the
embedded dashboard.
3. The AI agent backend connects to Superset via the **Superset MCP Server**
(`@modelcontextprotocol/server-superset` or native `/mcp/`), using tools like:
- `execute_sql`: validates and runs queries
- `list_datasets`: inspects table schemas
- `generate_explore_link`: generates pre-configured chart exploration URLs
4. When the AI generates a recommendation, it updates the embedded dashboard or
navigates the user directly to the generated explore link.
### Architecture 2: Frontend Script Injection via FLASK_APP_MUTATOR (In-App
Floating Widget)
If you want the chat window to appear directly inside the standalone Superset
UI without an external wrapper, you can inject a floating widget script via
`superset_config.py`:
```python
# In superset_config.py
def inject_chatbot_widget(app):
@app.after_request
def add_widget(response):
if response.content_type and "text/html" in response.content_type:
widget_code = '''
<script>
(function() {
var script = document.createElement("script");
script.src = "https://your-cdn.com/ai-chat-widget.js";;
script.async = true;
document.body.appendChild(script);
})();
</script>
'''
response.set_data(response.get_data().replace(b"</body>",
widget_code.encode("utf-8") + b"</body>"))
return response
FLASK_APP_MUTATOR = inject_chatbot_widget
```
Your `ai-chat-widget.js` script mounts a floating chat icon in the bottom-right
corner of the browser. When opened, it communicates with your LLM backend
orchestrating actions through Superset MCP or REST API endpoints
(`/api/v1/chart/data`, `/api/v1/dataset/`).
GitHub link:
https://github.com/apache/superset/discussions/41151#discussioncomment-19164795
----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
