Add reject mode to action of Moderation Extension API #1710

Closed
opened 2026-02-21 17:36:57 -05:00 by yindo · 0 comments
Owner

Originally created by @DatumWorld on GitHub (Apr 1, 2024).

Originally assigned to: @takatost, @VincePotato on GitHub.

Self Checks

  • I have searched for existing issues search for existing issues, including closed ones.
  • I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
  • Pleas do not modify this template :) and fill in all the required fields.

1. Is this request related to a challenge you're experiencing?

Some user input are too sensitive to be sent to the LLM outside,because they may contain Top Secret or confidential from company or organization,so,they should be rejected but not simply override。

2. Describe the feature you'd like to see

If Top Secret or confidential is found in user input,or rubbish and unhealthy info is found in output from LLMs,it should be rejected with some elegant tips。the field action should adds reject in Moderation Extension API on both input from user and output from the outside LLMs。

3. How will this feature improve your workflow or experience?

Preventing Top Secret or confidential to be leaked from inner company by someone casually,and Keeping the rubbish and unhealthy info outside

4. Additional context or comments

action=reject
{
"flagged": true,
"action": "direct_output",
"preset_response": "Your content violates our usage policy."
}

5. Can you help us with this feature?

  • I am interested in contributing to this feature.
Originally created by @DatumWorld on GitHub (Apr 1, 2024). Originally assigned to: @takatost, @VincePotato on GitHub. ### Self Checks - [X] I have searched for existing issues [search for existing issues](https://github.com/langgenius/dify/issues), including closed ones. - [X] I confirm that I am using English to submit this report (我已阅读并同意 [Language Policy](https://github.com/langgenius/dify/issues/1542)). - [X] Pleas do not modify this template :) and fill in all the required fields. ### 1. Is this request related to a challenge you're experiencing? Some user input are too sensitive to be sent to the LLM outside,because they may contain Top Secret or confidential from company or organization,so,they should be rejected but not simply override。 ### 2. Describe the feature you'd like to see If Top Secret or confidential is found in user input,or rubbish and unhealthy info is found in output from LLMs,it should be rejected with some elegant tips。the field action should adds reject in Moderation Extension API on both input from user and output from the outside LLMs。 ### 3. How will this feature improve your workflow or experience? Preventing Top Secret or confidential to be leaked from inner company by someone casually,and Keeping the rubbish and unhealthy info outside ### 4. Additional context or comments action=reject { "flagged": true, "action": "direct_output", "preset_response": "Your content violates our usage policy." } ### 5. Can you help us with this feature? - [X] I am interested in contributing to this feature.
yindo added the 💪 enhancementstale🌊 feat:workflow labels 2026-02-21 17:36:57 -05:00
yindo closed this issue 2026-02-21 17:36:58 -05:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langgenius/dify#1710