- DATA
- TUNING
- ANALYSIS
- HUMAN
Short answer
How can I send user ratings back to Dify automatically?
Your agent uses the Dify.AI SDK to trigger a feedback submission whenever a user provides a rating or comment. This creates a direct pipeline for RLHF and model tuning. You'll see how the data flows from the chat interface to your Dify dashboard.
Feedback outcomes
Where your data lands.
What happens after your agent captures a rating.
DATA
Logged feedback
The agent sends the rating and text to Dify. This populates your feedback logs for review.
TUNING
RLHF readiness
Captured data becomes available for Reinforcement Learning from Human Feedback. You use this to refine model responses.
ANALYSIS
Performance insights
Aggregated ratings show you exactly where the agent succeeds or fails. You get a clear view of model accuracy.
HUMAN
Manual review
You decide which feedback is high-quality enough to use for fine-tuning. The AI doesn't decide what to keep.
The workflow
What your AI does when the feedback arrives.
The agent acts as the bridge between the user's opinion and your Dify environment.
Capture input
The agent listens for specific user signals, like a thumbs up or a written critique of a response.
submit_feedbackProcess rating
The agent formats the user's rating and the associated conversation ID into a structured payload.
submit_feedbackSubmit to Dify
The agent pushes the formatted feedback directly into your Dify workspace using the provided tool.
submit_feedbackConfirm success
The agent verifies the submission was successful and can notify the user that their feedback was received.
submit_feedback
Try it
Copy these to start.
Use these prompts to trigger the feedback process.
Starting points
Swap out the specific chat IDs or user names to match your current session.
Dify.AI Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_esFwQvD1BTqmueeHwIOlGHPDJ8Oq0FzvLe2J4jkV/mcpClaude Desktop
Follow the steps below to connect in seconds.
- 1In Claude Desktop, open Settings → Connectors.
- 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
- 3Click Add and start a new chat — Dify.AI capabilities are ready to use.
{
"mcpServers": {
"difyai-sdk-mcp": {
"url": "https://edge.vinkius.com/vk_preview_esFwQvD1BTqmueeHwIOlGHPDJ8Oq0FzvLe2J4jkV/mcp"
}
}
}Submit a thumbs up feedback for the last response in this chat.
Send a rating of 1 to Dify for the response about the quarterly report.
Log this user comment as feedback for the current conversation.
Submit a negative rating for the last answer and include 'too wordy' as the reason.
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect Dify.AI SDK once, then ask.
One connection links your agent to the Dify API. You can find the full setup details on the Connector page.
Connect Dify.AI SDK to your AIFAQ
How this task behaves
- 01
Can the AI delete feedback once it is submitted?
No. The tool only allows for submitting new feedback to Dify.
- 02
Does the AI decide which ratings are important?
No. The AI simply submits whatever feedback the user provides. You must review the data in Dify to decide what to use for tuning.
- 03
Can I send feedback for a specific message ID?
Yes, you can instruct your agent to submit feedback for specific interactions.
- 04
Will this automatically retrain my model?
No. This tool only moves the data to Dify. You still need to initiate the training or tuning process yourself.
- 05
Can the agent submit feedback without user input?
No. The agent requires a user rating or comment to trigger the submission tool.
More questions about Dify.AI SDK? The Connector page answers them. See everything the Dify.AI SDK Connector can do
Connect Dify.AI to Claude, Cursor, ChatGPT & more
