Could you please remove the validate for "sys.query in user message is required" in llm node #5917

Closed
opened 2026-02-21 18:13:14 -05:00 by yindo · 12 comments
Owner

Originally created by @gaozhengzhou on GitHub (Oct 4, 2024).

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).
  • [FOR CHINESE USERS] 请务必使用英文提交 Issue,否则会被关闭。谢谢!:)
  • Please do not modify this template :) and fill in all the required fields.

1. Is this request related to a challenge you're experiencing? Tell me about your story.

Before calling the LLM, I parse the sys.query and pass the processed results to the LLM node. However, the LLM node checks that the user message must contain sys.query, which prevents me from doing this,so could you please remove the validate for "sys.query in user message is required" in llm node?
image

2. Additional context or comments

No response

3. Can you help us with this feature?

  • I am interested in contributing to this feature.
Originally created by @gaozhengzhou on GitHub (Oct 4, 2024). ### 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] [FOR CHINESE USERS] 请务必使用英文提交 Issue,否则会被关闭。谢谢!:) - [X] Please do not modify this template :) and fill in all the required fields. ### 1. Is this request related to a challenge you're experiencing? Tell me about your story. Before calling the LLM, I parse the sys.query and pass the processed results to the LLM node. However, the LLM node checks that the user message must contain sys.query, which prevents me from doing this,so could you please remove the validate for "sys.query in user message is required" in llm node? ![image](https://github.com/user-attachments/assets/8e683991-a48c-4258-a9e2-deb06fae33a8) ### 2. Additional context or comments _No response_ ### 3. Can you help us with this feature? - [X] I am interested in contributing to this feature.
yindo added the 💪 enhancement label 2026-02-21 18:13:14 -05:00
yindo closed this issue 2026-02-21 18:13:14 -05:00
Author
Owner

@crazywoola commented on GitHub (Oct 4, 2024):

If you turn on the memory, then it's required.

@crazywoola commented on GitHub (Oct 4, 2024): If you turn on the memory, then it's required.
Author
Owner

@sweetkey commented on GitHub (Nov 13, 2024):

Can this sys.query be changed to be optional? In a LLM block, the function is to summarize the input of the pre-define text at the start, regardless what the user enters as query (or stored as sys.query). But LLM block always consider the query as the primary text source. I don't need the input of sys.query for this perticular case.

@sweetkey commented on GitHub (Nov 13, 2024): Can this sys.query be changed to be optional? In a LLM block, the function is to summarize the input of the pre-define text at the start, regardless what the user enters as query (or stored as sys.query). But LLM block always consider the query as the primary text source. I don't need the input of sys.query for this perticular case.
Author
Owner

@aandis commented on GitHub (Nov 29, 2024):

+1 on this. Please implement this. The function of an LLM block is to summarise text from any preceding step or any static text. It does not necessarily have to be related to sys.query. This is a major blocker preventing us from going ahead with Dify in our company.

@aandis commented on GitHub (Nov 29, 2024): +1 on this. Please implement this. The function of an LLM block is to summarise text from any preceding step or any static text. It does not necessarily have to be related to `sys.query`. This is a major blocker preventing us from going ahead with Dify in our company.
Author
Owner

@nadirvishun commented on GitHub (Nov 29, 2024):

If I add a step LLM after the user's question to process the user's question, turn the user's question into similar multiple questions llm_result, then go through the knowledge base and another LLM, at that point what I need subsequently is not sys.query but llm_result.

@nadirvishun commented on GitHub (Nov 29, 2024): If I add a step LLM after the user's question to process the user's question, turn the user's question into similar multiple questions `llm_result`, then go through the knowledge base and another LLM, at that point what I need subsequently is not `sys.query` but `llm_result`.
Author
Owner

@00drdelius commented on GitHub (Dec 17, 2024):

+1. Strongly needed this feature.

@00drdelius commented on GitHub (Dec 17, 2024): +1. Strongly needed this feature.
Author
Owner

@wow-ww commented on GitHub (Jan 11, 2025):

+1. Strongly needed this feature.

@wow-ww commented on GitHub (Jan 11, 2025): +1. Strongly needed this feature.
Author
Owner

@zhangdazhuang commented on GitHub (Mar 5, 2025):

+1. Strongly needed this feature.
I use {{#sys.query#}} to make LLM ignore this message.

@zhangdazhuang commented on GitHub (Mar 5, 2025): +1. Strongly needed this feature. I use ~~{{#sys.query#}}~~ to make LLM ignore this message.
Author
Owner

@Silow9 commented on GitHub (Mar 19, 2025):

+1. Strongly needed this feature.

@Silow9 commented on GitHub (Mar 19, 2025): +1. Strongly needed this feature.
Author
Owner

@Sos-Zachary commented on GitHub (Jun 5, 2025):

+1. Strongly needed this feature.
i use a code_node to help me to preprocess the query string to get a well-structed input,but finally i also need to input sys.query,that is too inflexible. (

@Sos-Zachary commented on GitHub (Jun 5, 2025): +1. Strongly needed this feature. i use a code_node to help me to preprocess the query string to get a well-structed input,but finally i also need to input sys.query,that is too inflexible. (
Author
Owner

@allenhsu commented on GitHub (Jun 24, 2025):

In my case, the first block is using a local vision model to do OCR. The output of the OCR step will be used in the next block which calls a non-vision deep thinking model.

@allenhsu commented on GitHub (Jun 24, 2025): In my case, the first block is using a local vision model to do OCR. The output of the OCR step will be used in the next block which calls a non-vision deep thinking model.
Author
Owner

@MaoJianwei commented on GitHub (Nov 1, 2025):

+1. Strongly needed this feature.

@MaoJianwei commented on GitHub (Nov 1, 2025): +1. Strongly needed this feature.
Author
Owner

@Sagiri-12 commented on GitHub (Dec 4, 2025):

+1. Strongly needed this feature. Because I have already wrapper the query at certain nodes, but I am unable to use it.

@Sagiri-12 commented on GitHub (Dec 4, 2025): +1. Strongly needed this feature. Because I have already wrapper the query at certain nodes, but I am unable to use it.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langgenius/dify#5917