FoundationAgents/MetaGPT · error · InputRequired

prompt or messages is required!

Error message

prompt or messages is required!

What it means

DashScope generation entry (Generation.acall) requires either prompt or messages; if both are None/empty, InputRequired('prompt or messages is required!') is raised immediately, before any network call. It is a pure argument-validation guard.

Source

Thrown at metagpt/provider/dashscope_api.py:120

    request_data.add_parameters(**kwargs)
    request.data = request_data
    return request


class AGeneration(Generation, BaseAioApi):
    @classmethod
    async def acall(
        cls,
        model: str,
        prompt: Any = None,
        history: list = None,
        api_key: str = None,
        messages: List[Message] = None,
        plugins: Union[str, Dict[str, Any]] = None,
        **kwargs,
    ) -> Union[GenerationResponse, AsyncGenerator[GenerationResponse, None]]:
        if (prompt is None or not prompt) and (messages is None or not messages):
            raise InputRequired("prompt or messages is required!")
        if model is None or not model:
            raise ModelRequired("Model is required!")
        task_group, function = "aigc", "generation"  # fixed value
        if plugins is not None:
            headers = kwargs.pop("headers", {})
            if isinstance(plugins, str):
                headers["X-DashScope-Plugin"] = plugins
            else:
                headers["X-DashScope-Plugin"] = json.dumps(plugins)
            kwargs["headers"] = headers
        input, parameters = cls._build_input_parameters(model, prompt, history, messages, **kwargs)

        api_key, model = BaseAioApi._validate_params(api_key, model)
        request = build_api_arequest(
            model=model,
            input=input,
            task_group=task_group,
            task=Generation.task,

View on GitHub (pinned to 11cdf466d0)

Solutions

1. Pass a non-empty prompt or a non-empty messages list.
2. Add your own precondition check/log before calling so empty prompts are caught with better context.
3. If messages-based, ensure the messages list itself (not history) carries the conversation.

Example fix

// before
resp = await Generation.acall(model="qwen-plus", prompt="", messages=[])

// after
resp = await Generation.acall(model="qwen-plus", prompt="Summarize this document...")

def has_prompt(prompt, messages) -> bool:
    return bool(prompt) or bool(messages)

assert has_prompt(prompt, messages), "prompt or messages required"

try:
    resp = await Generation.acall(model=model, prompt=prompt, messages=messages)
except Exception as e:
    if "prompt or messages is required" in str(e):
        raise ValueError("refusing to call LLM with empty prompt") from e
    raise

When it happens

Trigger: Calling Generation.acall(model='qwen-plus') with neither prompt nor messages, or with prompt='' and messages=[]; template code where the prompt variable ended up empty.

Common situations: Building prompts dynamically and shipping an empty string; passing messages only as history (history alone does not satisfy the check); config-driven prompt templates resolving to empty.

Related errors

• Unsupported protocol: %s, support [http, https, websocket]
• There is no input data and form data
• Model is required!
• Only support for python, markdown, but got {language}
• Only support for language: python, markdown, but got {langua

AI-assisted analysis of FoundationAgents/MetaGPT@11cdf466d0 (2026-08-14). Data as JSON: /api/errors/5c8797d134587a2a. Report an issue: GitHub.