Occurring at least one message error when there was no user message for Claude llm on workflow app

langgenius / dify

Dify is an open-source LLM app development platform. Dify's intuitive interface combines AI workflow, RAG pipeline, agent capabilities, model management, observability features and more, letting you quickly go from prototype to production.

Other

51.81k stars 7.51k forks source link

Self Checks

[X] This is only for bug report, if you would like to ask a question, please head to Discussions.
[X] I have searched for existing issues search for existing 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).
[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.

Dify version

v0.8.3

Cloud or Self Hosted

Self Hosted (Source)

Steps to reproduce

created a workflow app with Calude LLM
only give a system message on LLM node
got error for at least one message is required
for other models, there's no issue for this case.

✔️ Expected Behavior

should no error and return the correct response, like this.

I already have a mr to mitigate this issue, pls check it out.

❌ Actual Behavior

got at least one message is required error.

I found a similar closed issue that might be relevant to your problem: bad_request: The browser (or proxy) sent a request that this server could not understand. The solution was to escape the newline character in the text ^[1].

Regarding the specific issue with Claude LLM requiring at least one user message, this is due to the implementation of the MinimaxChatCompletionPro class, which explicitly checks for the presence of a user message after processing the system message. If only a system message is provided, it raises a BadRequestError ^[2].

The relevant code snippet is as follows:

# check if there is a system message
if len(prompt_messages) == 0:
    raise BadRequestError("At least one message is required")
if prompt_messages[0].role == MinimaxMessage.Role.SYSTEM.value:
    if prompt_messages[0].content:
        bot_setting["content"] = prompt_messages[0].content
    prompt_messages = prompt_messages[1:]

# check if there is a user message
if len(prompt_messages) == 0:
    raise BadRequestError("At least one user message is required")

This requirement ensures proper interaction flow with the AI model. Other models might not have this constraint due to different implementation details.

Your pull request likely addresses this by ensuring that there is at least one user message and removing any empty system messages from the prompt messages list. This is a common approach to mitigate such issues ^[3].

To continue talking to Dosu, mention @dosu.

langgenius / dify