langgenius/dify
1
0
Fork 0
mirror of https://github.com/langgenius/dify.git synced 2026-07-22 12:36:14 -04:00
Code Issues 775 Packages Projects Releases 10 Wiki Activity

Oneapi with moonshot not work fine #4360

New Issue
Closed
opened 2026-02-21 18:06:04 -05:00 by yindo · 13 comments
yindo commented 2026-02-21 18:06:04 -05:00
Owner
Copy Link
Copy Source

Originally created by @hunter-eric on GitHub (Jul 1, 2024).

Self Checks

  • This is only for bug report, if you would like to ask a question, please head to Discussions.
  • 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).
  • 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.

Dify version

0.6.11

Cloud or Self Hosted

Self Hosted (Docker)

Steps to reproduce

I'm useing OneAPI to limit ai request frequency. when i using moonshot directly, function call with dify tool work perfectly.
but when using OneApi with Moonshot， it didn't work. I'm sure the option with "function call" in Oneapi with Moonshot is open!

✔️ Expected Behavior

using moonshot directly or using oneapi with moonshot should work same

❌ Actual Behavior

only using moonshot directly work fine

Originally created by @hunter-eric on GitHub (Jul 1, 2024). ### Self Checks - [X] This is only for bug report, if you would like to ask a question, please head to [Discussions](https://github.com/langgenius/dify/discussions/categories/general). - [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] 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 0.6.11 ### Cloud or Self Hosted Self Hosted (Docker) ### Steps to reproduce I'm useing OneAPI to limit ai request frequency. when i using moonshot directly, function call with dify tool work perfectly. but when using OneApi with Moonshot， it didn't work. I'm sure the option with "function call" in Oneapi with Moonshot is open! ### ✔️ Expected Behavior using moonshot directly or using oneapi with moonshot should work same ### ❌ Actual Behavior only using moonshot directly work fine
yindo closed this issue 2026-02-21 18:06:04 -05:00
yindo commented 2026-02-21 18:06:04 -05:00
Author
Owner
Copy Link
Copy Source

@dosubot[bot] commented on GitHub (Jul 1, 2024):

The issue with using OneAPI with Moonshot not working, while using Moonshot directly works fine in Dify version 0.6.11, is likely due to network accessibility problems. Specifically, if Dify and Moonshot are deployed using Docker, the localhost address usually refers to the container itself, not the host machine or other containers. This can cause connection issues.

To resolve this, you need to expose the Moonshot service to the network. Instead of using localhost, you should use host.docker.internal to refer to the Docker host machine.

http://host.docker.internal:11434

This change should allow OneAPI to access the Moonshot service correctly [1].

To continue talking to Dosu, mention @dosu.

@dosubot[bot] commented on GitHub (Jul 1, 2024): <!-- Answer --> The issue with using OneAPI with Moonshot not working, while using Moonshot directly works fine in Dify version 0.6.11, is likely due to network accessibility problems. Specifically, if Dify and Moonshot are deployed using Docker, the `localhost` address usually refers to the container itself, not the host machine or other containers. This can cause connection issues. To resolve this, you need to expose the Moonshot service to the network. Instead of using `localhost`, you should use `host.docker.internal` to refer to the Docker host machine. ```bash http://host.docker.internal:11434 ``` This change should allow OneAPI to access the Moonshot service correctly <sup>[[1]](https://github.com/langgenius/dify-docs/blob/main/en/guides/model-configuration/ollama.md)</sup>. <!-- Dosu Comment Footer --> *To continue talking to [Dosu](https://dosu.dev), mention @dosu.*
yindo commented 2026-02-21 18:06:04 -05:00
Author
Owner
Copy Link
Copy Source

@hunter-eric commented on GitHub (Jul 1, 2024):

The issue with using OneAPI with Moonshot not working, while using Moonshot directly works fine in Dify version 0.6.11, is likely due to network accessibility problems. Specifically, if Dify and Moonshot are deployed using Docker, the localhost address usually refers to the container itself, not the host machine or other containers. This can cause connection issues.

To resolve this, you need to expose the Moonshot service to the network. Instead of using localhost, you should use host.docker.internal to refer to the Docker host machine.

http://host.docker.internal:11434

This change should allow OneAPI to access the Moonshot service correctly [1].

To continue talking to Dosu, mention @dosu.

thanks for your reply. I'm sure Oneapi with moonshot is set correctly! when i use that just for chatting,it works fine. but when i use for dify tool ( like function call?),it not work!
WechatIMG3004

@hunter-eric commented on GitHub (Jul 1, 2024): > The issue with using OneAPI with Moonshot not working, while using Moonshot directly works fine in Dify version 0.6.11, is likely due to network accessibility problems. Specifically, if Dify and Moonshot are deployed using Docker, the `localhost` address usually refers to the container itself, not the host machine or other containers. This can cause connection issues. > > To resolve this, you need to expose the Moonshot service to the network. Instead of using `localhost`, you should use `host.docker.internal` to refer to the Docker host machine. > > ```shell > http://host.docker.internal:11434 > ``` > > This change should allow OneAPI to access the Moonshot service correctly [[1]](https://github.com/langgenius/dify-docs/blob/main/en/guides/model-configuration/ollama.md). > > _To continue talking to [Dosu](https://dosu.dev), mention @dosu._ thanks for your reply. I'm sure Oneapi with moonshot is set correctly! when i use that just for chatting,it works fine. but when i use for dify tool ( like function call?),it not work! ![WechatIMG3004](https://github.com/langgenius/dify/assets/1303119/ffdb04c8-bb07-4dba-ac3a-eb7d0412e21e)
yindo commented 2026-02-21 18:06:04 -05:00
Author
Owner
Copy Link
Copy Source

@dosubot[bot] commented on GitHub (Jul 1, 2024):

To resolve the issue where using OneAPI with Moonshot for function calls in Dify version 0.6.11 does not work correctly, you need to ensure that the function_calling_type in the credentials is set to 'tool_call'. This is necessary because the get_customizable_model_schema method checks this credential to determine if the model supports tool calls.

Here is the relevant part of the code:

def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
    return AIModelEntity(
        model=model,
        label=I18nObject(en_US=model, zh_Hans=model),
        model_type=ModelType.LLM,
        features=[ModelFeature.TOOL_CALL, ModelFeature.MULTI_TOOL_CALL, ModelFeature.STREAM_TOOL_CALL] 
            if credentials.get('function_calling_type') == 'tool_call' 
            else [],
        fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
        model_properties={
            ModelPropertyKey.CONTEXT_SIZE: int(credentials.get('context_size', 4096)),
            ModelPropertyKey.MODE: LLMMode.CHAT.value,
        },
        parameter_rules=[
            ParameterRule(
                name='temperature',
                use_template='temperature',
                label=I18nObject(en_US='Temperature', zh_Hans='温度'),
                type=ParameterType.FLOAT,
            ),
            ParameterRule(
                name='max_tokens',
                use_template='max_tokens',
                default=512,
                min=1,
                max=int(credentials.get('max_tokens', 4096)),
                label=I18nObject(en_US='Max Tokens', zh_Hans='最大标记'),
                type=ParameterType.INT,
            ),
            ParameterRule(
                name='top_p',
                use_template='top_p',
                label=I18nObject(en_US='Top P', zh_Hans='Top P'),
                type=ParameterType.FLOAT,
            ),
        ]
    )

Ensure that your credentials dictionary includes 'function_calling_type': 'tool_call' when invoking the model. Here is an example of how you might set up the credentials:

credentials = {
    'api_key': 'your_api_key',
    'function_calling_type': 'tool_call',
    'context_size': 4096,
    'max_tokens': 512,
    # other necessary credentials
}

By setting the function_calling_type to 'tool_call', the model will recognize that it supports tool calls, and the function calls should work correctly [1].

To continue talking to Dosu, mention @dosu.

@dosubot[bot] commented on GitHub (Jul 1, 2024): <!-- Answer --> To resolve the issue where using OneAPI with Moonshot for function calls in Dify version 0.6.11 does not work correctly, you need to ensure that the `function_calling_type` in the credentials is set to `'tool_call'`. This is necessary because the `get_customizable_model_schema` method checks this credential to determine if the model supports tool calls. Here is the relevant part of the code: ```python def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None: return AIModelEntity( model=model, label=I18nObject(en_US=model, zh_Hans=model), model_type=ModelType.LLM, features=[ModelFeature.TOOL_CALL, ModelFeature.MULTI_TOOL_CALL, ModelFeature.STREAM_TOOL_CALL] if credentials.get('function_calling_type') == 'tool_call' else [], fetch_from=FetchFrom.CUSTOMIZABLE_MODEL, model_properties={ ModelPropertyKey.CONTEXT_SIZE: int(credentials.get('context_size', 4096)), ModelPropertyKey.MODE: LLMMode.CHAT.value, }, parameter_rules=[ ParameterRule( name='temperature', use_template='temperature', label=I18nObject(en_US='Temperature', zh_Hans='温度'), type=ParameterType.FLOAT, ), ParameterRule( name='max_tokens', use_template='max_tokens', default=512, min=1, max=int(credentials.get('max_tokens', 4096)), label=I18nObject(en_US='Max Tokens', zh_Hans='最大标记'), type=ParameterType.INT, ), ParameterRule( name='top_p', use_template='top_p', label=I18nObject(en_US='Top P', zh_Hans='Top P'), type=ParameterType.FLOAT, ), ] ) ``` Ensure that your credentials dictionary includes `'function_calling_type': 'tool_call'` when invoking the model. Here is an example of how you might set up the credentials: ```python credentials = { 'api_key': 'your_api_key', 'function_calling_type': 'tool_call', 'context_size': 4096, 'max_tokens': 512, # other necessary credentials } ``` By setting the `function_calling_type` to `'tool_call'`, the model will recognize that it supports tool calls, and the function calls should work correctly <sup>[[1]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/moonshot/llm/llm.py)</sup>. <!-- Dosu Comment Footer --> *To continue talking to [Dosu](https://dosu.dev), mention @dosu.*
yindo commented 2026-02-21 18:06:04 -05:00
Author
Owner
Copy Link
Copy Source

@hunter-eric commented on GitHub (Jul 1, 2024):

works perfect! Thanks, dify team!

@hunter-eric commented on GitHub (Jul 1, 2024): works perfect! Thanks, dify team!
yindo commented 2026-02-21 18:06:04 -05:00
Author
Owner
Copy Link
Copy Source

@hunter-eric commented on GitHub (Jul 1, 2024):

sorry，it still have some problem with Oneapi（moonshot）. if use moonshot directly ,there's no problem.
WechatIMG3007

@hunter-eric commented on GitHub (Jul 1, 2024): sorry，it still have some problem with Oneapi（moonshot）. if use moonshot directly ,there's no problem. ![WechatIMG3007](https://github.com/langgenius/dify/assets/1303119/91b80df1-6cd6-438d-8cbc-eb329f05bcc2)
yindo commented 2026-02-21 18:06:05 -05:00
Author
Owner
Copy Link
Copy Source

@hunter-eric commented on GitHub (Jul 1, 2024):

sorry，it still have some problem with Oneapi（moonshot）. if use moonshot directly ,there's no problem. WechatIMG3007

seems like lost some params when using oneapi with ToolCall

@hunter-eric commented on GitHub (Jul 1, 2024): > sorry，it still have some problem with Oneapi（moonshot）. if use moonshot directly ,there's no problem. ![WechatIMG3007](https://private-user-images.githubusercontent.com/1303119/344650462-91b80df1-6cd6-438d-8cbc-eb329f05bcc2.jpg?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.Bup5mBki3lvrZGzKATnrIY49RLszpV5jQZnZRZ25wcU) seems like lost some params when using oneapi with ToolCall
yindo commented 2026-02-21 18:06:05 -05:00
Author
Owner
Copy Link
Copy Source

@dosubot[bot] commented on GitHub (Jul 1, 2024):

The "Invalid request: agent message content is empty" error suggests that the prompt_messages parameter might be empty or not properly populated when using OneAPI with Moonshot. Ensure that the prompt_messages list contains valid PromptMessage objects before invoking the method.

Here is an example of how to set up the prompt_messages:

from core.model_runtime.entities.message_entities import UserPromptMessage, AssistantPromptMessage

prompt_messages = [
    UserPromptMessage(content="Your user prompt here"),
    AssistantPromptMessage(content="Your assistant prompt here")
]

credentials = {
    'api_key': 'your_api_key',
    'function_calling_type': 'tool_call',
    'context_size': 4096,
    'max_tokens': 512,
    # other necessary credentials
}

model_parameters = {
    'temperature': 0.7,
    'top_p': 0.9,
    'max_tokens': 512
}

# Assuming `moonshot_model` is an instance of `MoonshotLargeLanguageModel`
result = moonshot_model._invoke(
    model="moonshot-v1-8k",
    credentials=credentials,
    prompt_messages=prompt_messages,
    model_parameters=model_parameters
)

Ensure that the prompt_messages list is not empty and contains the necessary message content. This should help resolve the "Invalid request: agent message content is empty" error [1].

To continue talking to Dosu, mention @dosu.

@dosubot[bot] commented on GitHub (Jul 1, 2024): <!-- Answer --> The "Invalid request: agent message content is empty" error suggests that the `prompt_messages` parameter might be empty or not properly populated when using OneAPI with Moonshot. Ensure that the `prompt_messages` list contains valid `PromptMessage` objects before invoking the method. Here is an example of how to set up the `prompt_messages`: ```python from core.model_runtime.entities.message_entities import UserPromptMessage, AssistantPromptMessage prompt_messages = [ UserPromptMessage(content="Your user prompt here"), AssistantPromptMessage(content="Your assistant prompt here") ] credentials = { 'api_key': 'your_api_key', 'function_calling_type': 'tool_call', 'context_size': 4096, 'max_tokens': 512, # other necessary credentials } model_parameters = { 'temperature': 0.7, 'top_p': 0.9, 'max_tokens': 512 } # Assuming `moonshot_model` is an instance of `MoonshotLargeLanguageModel` result = moonshot_model._invoke( model="moonshot-v1-8k", credentials=credentials, prompt_messages=prompt_messages, model_parameters=model_parameters ) ``` Ensure that the `prompt_messages` list is not empty and contains the necessary message content. This should help resolve the "Invalid request: agent message content is empty" error <sup>[[1]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/moonshot/llm/llm.py)</sup>. <!-- Dosu Comment Footer --> *To continue talking to [Dosu](https://dosu.dev), mention @dosu.*
yindo commented 2026-02-21 18:06:05 -05:00
Author
Owner
Copy Link
Copy Source

@hunter-eric commented on GitHub (Jul 1, 2024):

The "Invalid request: agent message content is empty" error suggests that the prompt_messages parameter might be empty or not properly populated when using OneAPI with Moonshot. Ensure that the prompt_messages list contains valid PromptMessage objects before invoking the method.

Here is an example of how to set up the prompt_messages:

from core.model_runtime.entities.message_entities import UserPromptMessage, AssistantPromptMessage

prompt_messages = [
    UserPromptMessage(content="Your user prompt here"),
    AssistantPromptMessage(content="Your assistant prompt here")
]

credentials = {
    'api_key': 'your_api_key',
    'function_calling_type': 'tool_call',
    'context_size': 4096,
    'max_tokens': 512,
    # other necessary credentials
}

model_parameters = {
    'temperature': 0.7,
    'top_p': 0.9,
    'max_tokens': 512
}

# Assuming `moonshot_model` is an instance of `MoonshotLargeLanguageModel`
result = moonshot_model._invoke(
    model="moonshot-v1-8k",
    credentials=credentials,
    prompt_messages=prompt_messages,
    model_parameters=model_parameters
)

Ensure that the prompt_messages list is not empty and contains the necessary message content. This should help resolve the "Invalid request: agent message content is empty" error [1].

To continue talking to Dosu, mention @dosu.

I'm using dify web chat window. Not using api. the response is from dify.
by the way, Toocall response is good from Logs. image

@hunter-eric commented on GitHub (Jul 1, 2024): > The "Invalid request: agent message content is empty" error suggests that the `prompt_messages` parameter might be empty or not properly populated when using OneAPI with Moonshot. Ensure that the `prompt_messages` list contains valid `PromptMessage` objects before invoking the method. > > Here is an example of how to set up the `prompt_messages`: > > ```python > from core.model_runtime.entities.message_entities import UserPromptMessage, AssistantPromptMessage > > prompt_messages = [ > UserPromptMessage(content="Your user prompt here"), > AssistantPromptMessage(content="Your assistant prompt here") > ] > > credentials = { > 'api_key': 'your_api_key', > 'function_calling_type': 'tool_call', > 'context_size': 4096, > 'max_tokens': 512, > # other necessary credentials > } > > model_parameters = { > 'temperature': 0.7, > 'top_p': 0.9, > 'max_tokens': 512 > } > > # Assuming `moonshot_model` is an instance of `MoonshotLargeLanguageModel` > result = moonshot_model._invoke( > model="moonshot-v1-8k", > credentials=credentials, > prompt_messages=prompt_messages, > model_parameters=model_parameters > ) > ``` > > Ensure that the `prompt_messages` list is not empty and contains the necessary message content. This should help resolve the "Invalid request: agent message content is empty" error [[1]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/moonshot/llm/llm.py). > > _To continue talking to [Dosu](https://dosu.dev), mention @dosu._ I'm using dify web chat window. Not using api. the response is from dify. by the way, Toocall response is good from Logs. <img width="1312" alt="image" src="https://github.com/langgenius/dify/assets/1303119/b699595e-cc6c-4988-a77c-343a18815cde">
yindo commented 2026-02-21 18:06:05 -05:00
Author
Owner
Copy Link
Copy Source

@hunter-eric commented on GitHub (Jul 1, 2024):

dify should send tool response with context to oneapi(moonshot). But Oneapi response with 400(no params)
seem like interrupt by some error

@hunter-eric commented on GitHub (Jul 1, 2024): dify should send tool response with context to oneapi(moonshot). But Oneapi response with 400(no params) seem like interrupt by some error
yindo commented 2026-02-21 18:06:05 -05:00
Author
Owner
Copy Link
Copy Source

@hunter-eric commented on GitHub (Jul 3, 2024):

I'v seen this issue is marked as [feat:model-runtime],what does that mean? should i wait for latest release?

@hunter-eric commented on GitHub (Jul 3, 2024): I'v seen this issue is marked as [feat:model-runtime],what does that mean? should i wait for latest release?
yindo commented 2026-02-21 18:06:05 -05:00
Author
Owner
Copy Link
Copy Source

@dosubot[bot] commented on GitHub (Jul 3, 2024):

The issue you're experiencing with the "Invalid request: agent message content is empty" error when using OneAPI with Moonshot in Dify version 0.6.11 is likely related to the prompt_messages parameter not being properly populated or formatted.

Ensure that your prompt_messages list contains valid PromptMessage objects and that they are correctly converted to the required dictionary format. Here is a snippet from the code that converts PromptMessage objects to the required dictionary format:

def _convert_prompt_message_to_dict(self, message: PromptMessage, credentials: Optional[dict] = None) -> dict:
    if isinstance(message, UserPromptMessage):
        message = cast(UserPromptMessage, message)
        if isinstance(message.content, str):
            message_dict = {"role": "user", "content": message.content}
        else:
            sub_messages = []
            for message_content in message.content:
                if message_content.type == PromptMessageContentType.TEXT:
                    message_content = cast(PromptMessageContent, message_content)
                    sub_message_dict = {
                        "type": "text",
                        "text": message_content.data
                    }
                    sub_messages.append(sub_message_dict)
                elif message_content.type == PromptMessageContentType.IMAGE:
                    message_content = cast(ImagePromptMessageContent, message_content)
                    sub_message_dict = {
                        "type": "image_url",
                        "image_url": {
                            "url": message_content.data,
                            "detail": message_content.detail.value
                        }
                    }
                    sub_messages.append(sub_message_dict)
            message_dict = {"role": "user", "content": sub_messages}
    elif isinstance(message, AssistantPromptMessage):
        message = cast(AssistantPromptMessage, message)
        message_dict = {"role": "assistant", "content": message.content}
        if message.tool_calls:
            message_dict["tool_calls"] = []
            for function_call in message.tool_calls:
                message_dict["tool_calls"].append({
                    "id": function_call.id,
                    "type": function_call.type,
                    "function": {
                        "name": function_call.function.name,
                        "arguments": function_call.function.arguments
                    }
                })
    elif isinstance(message, ToolPromptMessage):
        message = cast(ToolPromptMessage, message)
        message_dict = {"role": "tool", "content": message.content, "tool_call_id": message.tool_call_id}
    elif isinstance(message, SystemPromptMessage):
        message = cast(SystemPromptMessage, message)
        message_dict = {"role": "system", "content": message.content}
    else:
        raise ValueError(f"Got unknown type {message}")
    if message.name:
        message_dict["name"] = message.name

    return message_dict

Ensure that your prompt_messages are correctly formatted and populated before making the API call. The PromptMessage class and its subclasses (e.g., UserPromptMessage, AssistantPromptMessage, SystemPromptMessage, ToolPromptMessage) should be properly instantiated with the required attributes [1][2].

Regarding the [feat:model-runtime] label, it indicates that the issue is related to the model runtime feature. There have been recent updates in the latest version (v0.6.12-fix1) that address issues related to prompt_messages and the feat:model-runtime feature. These updates ensure that the handling of prompt_messages and the construction of parameters for the glm-4v model are more robust [3][4].

If the issue persists, it might be related to the specific implementation of the OneAPI with Moonshot, and you may need to wait for the latest release to resolve this issue, especially since it is marked as [feat:model-runtime].

To continue talking to Dosu, mention @dosu.

@dosubot[bot] commented on GitHub (Jul 3, 2024): <!-- Answer --> The issue you're experiencing with the "Invalid request: agent message content is empty" error when using OneAPI with Moonshot in Dify version 0.6.11 is likely related to the `prompt_messages` parameter not being properly populated or formatted. Ensure that your `prompt_messages` list contains valid `PromptMessage` objects and that they are correctly converted to the required dictionary format. Here is a snippet from the code that converts `PromptMessage` objects to the required dictionary format: ```python def _convert_prompt_message_to_dict(self, message: PromptMessage, credentials: Optional[dict] = None) -> dict: if isinstance(message, UserPromptMessage): message = cast(UserPromptMessage, message) if isinstance(message.content, str): message_dict = {"role": "user", "content": message.content} else: sub_messages = [] for message_content in message.content: if message_content.type == PromptMessageContentType.TEXT: message_content = cast(PromptMessageContent, message_content) sub_message_dict = { "type": "text", "text": message_content.data } sub_messages.append(sub_message_dict) elif message_content.type == PromptMessageContentType.IMAGE: message_content = cast(ImagePromptMessageContent, message_content) sub_message_dict = { "type": "image_url", "image_url": { "url": message_content.data, "detail": message_content.detail.value } } sub_messages.append(sub_message_dict) message_dict = {"role": "user", "content": sub_messages} elif isinstance(message, AssistantPromptMessage): message = cast(AssistantPromptMessage, message) message_dict = {"role": "assistant", "content": message.content} if message.tool_calls: message_dict["tool_calls"] = [] for function_call in message.tool_calls: message_dict["tool_calls"].append({ "id": function_call.id, "type": function_call.type, "function": { "name": function_call.function.name, "arguments": function_call.function.arguments } }) elif isinstance(message, ToolPromptMessage): message = cast(ToolPromptMessage, message) message_dict = {"role": "tool", "content": message.content, "tool_call_id": message.tool_call_id} elif isinstance(message, SystemPromptMessage): message = cast(SystemPromptMessage, message) message_dict = {"role": "system", "content": message.content} else: raise ValueError(f"Got unknown type {message}") if message.name: message_dict["name"] = message.name return message_dict ``` Ensure that your `prompt_messages` are correctly formatted and populated before making the API call. The `PromptMessage` class and its subclasses (e.g., `UserPromptMessage`, `AssistantPromptMessage`, `SystemPromptMessage`, `ToolPromptMessage`) should be properly instantiated with the required attributes <sup>[[1]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/moonshot/llm/llm.py)</sup><sup>[[2]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/entities/message_entities.py)</sup>. Regarding the [feat:model-runtime] label, it indicates that the issue is related to the model runtime feature. There have been recent updates in the latest version (v0.6.12-fix1) that address issues related to `prompt_messages` and the `feat:model-runtime` feature. These updates ensure that the handling of `prompt_messages` and the construction of parameters for the `glm-4v` model are more robust <sup>[[3]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/zhipuai/llm/llm.py)</sup><sup>[[4]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/anthropic/llm/llm.py)</sup>. If the issue persists, it might be related to the specific implementation of the OneAPI with Moonshot, and you may need to wait for the latest release to resolve this issue, especially since it is marked as [feat:model-runtime]. <!-- Dosu Comment Footer --> *To continue talking to [Dosu](https://dosu.dev), mention @dosu.*
yindo commented 2026-02-21 18:06:05 -05:00
Author
Owner
Copy Link
Copy Source

@hunter-eric commented on GitHub (Jul 3, 2024):

The issue you're experiencing with the "Invalid request: agent message content is empty" error when using OneAPI with Moonshot in Dify version 0.6.11 is likely related to the prompt_messages parameter not being properly populated or formatted.

Ensure that your prompt_messages list contains valid PromptMessage objects and that they are correctly converted to the required dictionary format. Here is a snippet from the code that converts PromptMessage objects to the required dictionary format:

def _convert_prompt_message_to_dict(self, message: PromptMessage, credentials: Optional[dict] = None) -> dict:
    if isinstance(message, UserPromptMessage):
        message = cast(UserPromptMessage, message)
        if isinstance(message.content, str):
            message_dict = {"role": "user", "content": message.content}
        else:
            sub_messages = []
            for message_content in message.content:
                if message_content.type == PromptMessageContentType.TEXT:
                    message_content = cast(PromptMessageContent, message_content)
                    sub_message_dict = {
                        "type": "text",
                        "text": message_content.data
                    }
                    sub_messages.append(sub_message_dict)
                elif message_content.type == PromptMessageContentType.IMAGE:
                    message_content = cast(ImagePromptMessageContent, message_content)
                    sub_message_dict = {
                        "type": "image_url",
                        "image_url": {
                            "url": message_content.data,
                            "detail": message_content.detail.value
                        }
                    }
                    sub_messages.append(sub_message_dict)
            message_dict = {"role": "user", "content": sub_messages}
    elif isinstance(message, AssistantPromptMessage):
        message = cast(AssistantPromptMessage, message)
        message_dict = {"role": "assistant", "content": message.content}
        if message.tool_calls:
            message_dict["tool_calls"] = []
            for function_call in message.tool_calls:
                message_dict["tool_calls"].append({
                    "id": function_call.id,
                    "type": function_call.type,
                    "function": {
                        "name": function_call.function.name,
                        "arguments": function_call.function.arguments
                    }
                })
    elif isinstance(message, ToolPromptMessage):
        message = cast(ToolPromptMessage, message)
        message_dict = {"role": "tool", "content": message.content, "tool_call_id": message.tool_call_id}
    elif isinstance(message, SystemPromptMessage):
        message = cast(SystemPromptMessage, message)
        message_dict = {"role": "system", "content": message.content}
    else:
        raise ValueError(f"Got unknown type {message}")
    if message.name:
        message_dict["name"] = message.name

    return message_dict

Ensure that your prompt_messages are correctly formatted and populated before making the API call. The PromptMessage class and its subclasses (e.g., UserPromptMessage, AssistantPromptMessage, SystemPromptMessage, ToolPromptMessage) should be properly instantiated with the required attributes [1][2].

Regarding the [feat:model-runtime] label, it indicates that the issue is related to the model runtime feature. There have been recent updates in the latest version (v0.6.12-fix1) that address issues related to prompt_messages and the feat:model-runtime feature. These updates ensure that the handling of prompt_messages and the construction of parameters for the glm-4v model are more robust [3][4].

If the issue persists, it might be related to the specific implementation of the OneAPI with Moonshot, and you may need to wait for the latest release to resolve this issue, especially since it is marked as [feat:model-runtime].

To continue talking to Dosu, mention @dosu.

after upgrade to the latest version(v0.6.12-fix1)，it gave another error "lnvalid reauest: tool call id is not found"
image

@hunter-eric commented on GitHub (Jul 3, 2024): > The issue you're experiencing with the "Invalid request: agent message content is empty" error when using OneAPI with Moonshot in Dify version 0.6.11 is likely related to the `prompt_messages` parameter not being properly populated or formatted. > > Ensure that your `prompt_messages` list contains valid `PromptMessage` objects and that they are correctly converted to the required dictionary format. Here is a snippet from the code that converts `PromptMessage` objects to the required dictionary format: > > ```python > def _convert_prompt_message_to_dict(self, message: PromptMessage, credentials: Optional[dict] = None) -> dict: > if isinstance(message, UserPromptMessage): > message = cast(UserPromptMessage, message) > if isinstance(message.content, str): > message_dict = {"role": "user", "content": message.content} > else: > sub_messages = [] > for message_content in message.content: > if message_content.type == PromptMessageContentType.TEXT: > message_content = cast(PromptMessageContent, message_content) > sub_message_dict = { > "type": "text", > "text": message_content.data > } > sub_messages.append(sub_message_dict) > elif message_content.type == PromptMessageContentType.IMAGE: > message_content = cast(ImagePromptMessageContent, message_content) > sub_message_dict = { > "type": "image_url", > "image_url": { > "url": message_content.data, > "detail": message_content.detail.value > } > } > sub_messages.append(sub_message_dict) > message_dict = {"role": "user", "content": sub_messages} > elif isinstance(message, AssistantPromptMessage): > message = cast(AssistantPromptMessage, message) > message_dict = {"role": "assistant", "content": message.content} > if message.tool_calls: > message_dict["tool_calls"] = [] > for function_call in message.tool_calls: > message_dict["tool_calls"].append({ > "id": function_call.id, > "type": function_call.type, > "function": { > "name": function_call.function.name, > "arguments": function_call.function.arguments > } > }) > elif isinstance(message, ToolPromptMessage): > message = cast(ToolPromptMessage, message) > message_dict = {"role": "tool", "content": message.content, "tool_call_id": message.tool_call_id} > elif isinstance(message, SystemPromptMessage): > message = cast(SystemPromptMessage, message) > message_dict = {"role": "system", "content": message.content} > else: > raise ValueError(f"Got unknown type {message}") > if message.name: > message_dict["name"] = message.name > > return message_dict > ``` > > Ensure that your `prompt_messages` are correctly formatted and populated before making the API call. The `PromptMessage` class and its subclasses (e.g., `UserPromptMessage`, `AssistantPromptMessage`, `SystemPromptMessage`, `ToolPromptMessage`) should be properly instantiated with the required attributes [[1]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/moonshot/llm/llm.py)[[2]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/entities/message_entities.py). > > Regarding the [feat:model-runtime] label, it indicates that the issue is related to the model runtime feature. There have been recent updates in the latest version (v0.6.12-fix1) that address issues related to `prompt_messages` and the `feat:model-runtime` feature. These updates ensure that the handling of `prompt_messages` and the construction of parameters for the `glm-4v` model are more robust [[3]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/zhipuai/llm/llm.py)[[4]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/anthropic/llm/llm.py). > > If the issue persists, it might be related to the specific implementation of the OneAPI with Moonshot, and you may need to wait for the latest release to resolve this issue, especially since it is marked as [feat:model-runtime]. > > _To continue talking to [Dosu](https://dosu.dev), mention @dosu._ after upgrade to the latest version(v0.6.12-fix1)，it gave another error "lnvalid reauest: tool call id is not found" <img width="1025" alt="image" src="https://github.com/langgenius/dify/assets/1303119/8b6e41fe-c420-4a39-84e3-1385de3ed92b">
yindo commented 2026-02-21 18:06:05 -05:00
Author
Owner
Copy Link
Copy Source

@dosubot[bot] commented on GitHub (Jul 3, 2024):

To resolve the "Invalid request: tool_call_id is not found" error after upgrading to Dify version 0.6.12-fix1, ensure that the tool_call_id is correctly set in the ToolPromptMessage objects within your prompt_messages list. The tool_call_id must match the id of the tool call in the AssistantPromptMessage.

Here is a relevant snippet from the code that shows how tool_call_id is used:

def _convert_prompt_messages_to_message_and_chat_histories(self, prompt_messages: list[PromptMessage]) \
            -> tuple[str, list[ChatMessage], list[ChatStreamRequestToolResultsItem]]:
        chat_histories = []
        latest_tool_call_n_outputs = []
        for prompt_message in prompt_messages:
            if prompt_message.role == PromptMessageRole.ASSISTANT:
                prompt_message = cast(AssistantPromptMessage, prompt_message)
                if prompt_message.tool_calls:
                    for tool_call in prompt_message.tool_calls:
                        latest_tool_call_n_outputs.append(ChatStreamRequestToolResultsItem(
                            call=ToolCall(
                                name=tool_call.function.name,
                                parameters=json.loads(tool_call.function.arguments)
                            ),
                            outputs=[]
                        ))
                else:
                    cohere_prompt_message = self._convert_prompt_message_to_dict(prompt_message)
                    if cohere_prompt_message:
                        chat_histories.append(cohere_prompt_message)
            elif prompt_message.role == PromptMessageRole.TOOL:
                prompt_message = cast(ToolPromptMessage, prompt_message)
                if latest_tool_call_n_outputs:
                    i = 0
                    for tool_call_n_outputs in latest_tool_call_n_outputs:
                        if tool_call_n_outputs.call.name == prompt_message.tool_call_id:
                            latest_tool_call_n_outputs[i] = ChatStreamRequestToolResultsItem(
                                call=ToolCall(
                                    name=tool_call_n_outputs.call.name,
                                    parameters=tool_call_n_outputs.call.parameters
                                ),
                                outputs=[{
                                    "result": prompt_message.content
                                }]
                            )
                            break
                        i += 1
            else:
                cohere_prompt_message = self._convert_prompt_message_to_dict(prompt_message)
                if cohere_prompt_message:
                    chat_histories.append(cohere_prompt_message)
        if latest_tool_call_n_outputs:
            new_latest_tool_call_n_outputs = []
            for tool_call_n_outputs in latest_tool_call_n_outputs:
                if tool_call_n_outputs.outputs:
                    new_latest_tool_call_n_outputs.append(tool_call_n_outputs)

            latest_tool_call_n_outputs = new_latest_tool_call_n_outputs

        if len(chat_histories) > 0:
            latest_message = chat_histories.pop()
            message = latest_message.message
        else:
            raise ValueError('Prompt messages is empty')

        return message, chat_histories, latest_tool_call_n_outputs

Additionally, ensure that the tool_call_id is included in the parameters when invoking the model, as shown in the _chat_generate method:

def _chat_generate(self, model: str, credentials: dict,
                   prompt_messages: list[PromptMessage], model_parameters: dict,
                   tools: Optional[list[PromptMessageTool]] = None, stop: Optional[list[str]] = None,
                   stream: bool = True, user: Optional[str] = None) -> Union[LLMResult, Generator]:

    client = AzureOpenAI(**self._to_credential_kwargs(credentials))

    response_format = model_parameters.get("response_format")
    if response_format:
        if response_format == "json_object":
            response_format = {"type": "json_object"}
        else:
            response_format = {"type": "text"}

        model_parameters["response_format"] = response_format

    extra_model_kwargs = {}

    if tools:
        extra_model_kwargs['tools'] = [helper.dump_model(PromptMessageFunction(function=tool)) for tool in tools]

    if stop:
        extra_model_kwargs['stop'] = stop

    if user:
        extra_model_kwargs['user'] = user

    messages = [self._convert_prompt_message_to_dict(m) for m in prompt_messages]
    response = client.chat.completions.create(
        messages=messages,
        model=model,
        stream=stream,
        **model_parameters,
        **extra_model_kwargs,
    )

    if stream:
        return self._handle_chat_generate_stream_response(model, credentials, response, prompt_messages, tools)

    return self._handle_chat_generate_response(model, credentials, response, prompt_messages, tools)

Ensure that each ToolPromptMessage has a valid tool_call_id that corresponds to an existing tool call in the AssistantPromptMessage [1][2][3][4][5].

To continue talking to Dosu, mention @dosu.

@dosubot[bot] commented on GitHub (Jul 3, 2024): <!-- Answer --> To resolve the "Invalid request: tool_call_id is not found" error after upgrading to Dify version 0.6.12-fix1, ensure that the `tool_call_id` is correctly set in the `ToolPromptMessage` objects within your `prompt_messages` list. The `tool_call_id` must match the `id` of the tool call in the `AssistantPromptMessage`. Here is a relevant snippet from the code that shows how `tool_call_id` is used: ```python def _convert_prompt_messages_to_message_and_chat_histories(self, prompt_messages: list[PromptMessage]) \ -> tuple[str, list[ChatMessage], list[ChatStreamRequestToolResultsItem]]: chat_histories = [] latest_tool_call_n_outputs = [] for prompt_message in prompt_messages: if prompt_message.role == PromptMessageRole.ASSISTANT: prompt_message = cast(AssistantPromptMessage, prompt_message) if prompt_message.tool_calls: for tool_call in prompt_message.tool_calls: latest_tool_call_n_outputs.append(ChatStreamRequestToolResultsItem( call=ToolCall( name=tool_call.function.name, parameters=json.loads(tool_call.function.arguments) ), outputs=[] )) else: cohere_prompt_message = self._convert_prompt_message_to_dict(prompt_message) if cohere_prompt_message: chat_histories.append(cohere_prompt_message) elif prompt_message.role == PromptMessageRole.TOOL: prompt_message = cast(ToolPromptMessage, prompt_message) if latest_tool_call_n_outputs: i = 0 for tool_call_n_outputs in latest_tool_call_n_outputs: if tool_call_n_outputs.call.name == prompt_message.tool_call_id: latest_tool_call_n_outputs[i] = ChatStreamRequestToolResultsItem( call=ToolCall( name=tool_call_n_outputs.call.name, parameters=tool_call_n_outputs.call.parameters ), outputs=[{ "result": prompt_message.content }] ) break i += 1 else: cohere_prompt_message = self._convert_prompt_message_to_dict(prompt_message) if cohere_prompt_message: chat_histories.append(cohere_prompt_message) if latest_tool_call_n_outputs: new_latest_tool_call_n_outputs = [] for tool_call_n_outputs in latest_tool_call_n_outputs: if tool_call_n_outputs.outputs: new_latest_tool_call_n_outputs.append(tool_call_n_outputs) latest_tool_call_n_outputs = new_latest_tool_call_n_outputs if len(chat_histories) > 0: latest_message = chat_histories.pop() message = latest_message.message else: raise ValueError('Prompt messages is empty') return message, chat_histories, latest_tool_call_n_outputs ``` Additionally, ensure that the `tool_call_id` is included in the parameters when invoking the model, as shown in the `_chat_generate` method: ```python def _chat_generate(self, model: str, credentials: dict, prompt_messages: list[PromptMessage], model_parameters: dict, tools: Optional[list[PromptMessageTool]] = None, stop: Optional[list[str]] = None, stream: bool = True, user: Optional[str] = None) -> Union[LLMResult, Generator]: client = AzureOpenAI(**self._to_credential_kwargs(credentials)) response_format = model_parameters.get("response_format") if response_format: if response_format == "json_object": response_format = {"type": "json_object"} else: response_format = {"type": "text"} model_parameters["response_format"] = response_format extra_model_kwargs = {} if tools: extra_model_kwargs['tools'] = [helper.dump_model(PromptMessageFunction(function=tool)) for tool in tools] if stop: extra_model_kwargs['stop'] = stop if user: extra_model_kwargs['user'] = user messages = [self._convert_prompt_message_to_dict(m) for m in prompt_messages] response = client.chat.completions.create( messages=messages, model=model, stream=stream, **model_parameters, **extra_model_kwargs, ) if stream: return self._handle_chat_generate_stream_response(model, credentials, response, prompt_messages, tools) return self._handle_chat_generate_response(model, credentials, response, prompt_messages, tools) ``` Ensure that each `ToolPromptMessage` has a valid `tool_call_id` that corresponds to an existing tool call in the `AssistantPromptMessage` <sup>[[1]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/anthropic/llm/llm.py)</sup><sup>[[2]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/xinference/llm/llm.py)</sup><sup>[[3]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/cohere/llm/llm.py)</sup><sup>[[4]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/zhipuai/llm/llm.py)</sup><sup>[[5]](https://github.com/langgenius/dify/blob/main/api/core/model_runtime/model_providers/azure_openai/llm/llm.py)</sup>. <!-- Dosu Comment Footer --> *To continue talking to [Dosu](https://dosu.dev), mention @dosu.*
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
Knowledge Pipeline
Proposal
WIP
cloud
dependencies
feat:webapp
good first issue
outdated
performance
pull-request
Mirrored from GitHub Pull Request
 python:uv
refactor
size:L
stale
status: accepting prs
status: blocked
status: in discussion
stt & tts
vulnerablity
web
wont-fix
🌊 feat:workflow
🌚 invalid
🐞 bug
👻 feat:rag
💪 enhancement
📚 documentation
🔨 feat:tools
🙋‍♂️ question
🙏 help wanted
🤔 cant-reproduce
🤖 feat:agent
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langgenius/dify#4360
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?