agentscope-ai/agentscope · error · ValueError

Each schema must be a dict with 'type' as 'function' and 'fu

Error message

Each schema must be a dict with 'type' as 'function' and 'function' key, got {value}

What it means

The DashScope model adapter requires each tool schema to be an OpenAI-style dict with type=='function' and a 'function' key. Passing raw function dicts (missing the wrapper), malformed schemas, or non-dict entries raises ValueError in _format_tools.

Source

Thrown at src/agentscope/model/_dashscope/_model.py:516

            `tuple[list[dict] | None, str | dict | None]`:
                A tuple of (formatted_tools, formatted_tool_choice).
        """
        if tool_choice and tools:
            self._validate_tool_choice(tool_choice, tools)
            if tool_choice.tools:
                allowed = set(tool_choice.tools)
                tools = [t for t in tools if t["function"]["name"] in allowed]

        fmt_tools = None
        if tools:
            for value in tools:
                if (
                    not isinstance(value, dict)
                    or "type" not in value
                    or value["type"] != "function"
                    or "function" not in value
                ):
                    raise ValueError(
                        f"Each schema must be a dict with 'type' as "
                        f"'function' and 'function' key, got {value}",
                    )
            fmt_tools = tools

        if not tool_choice:
            return fmt_tools, None

        mode = tool_choice.mode

        if mode not in _TOOL_CHOICE_LITERAL_MODES:
            return fmt_tools, {
                "type": "function",
                "function": {"name": mode},
            }

        if mode == "required":
            warnings.warn(

View on GitHub (pinned to e90f1c7592)

Solutions

1. Wrap each tool in the OpenAI function envelope: {'type': 'function', 'function': {'name': ..., 'description': ..., 'parameters': ...}}
2. Prefer registering tools via agentscope Toolkit so schemas are generated in the correct format
3. Validate each entry with the check shown in the error before submitting

Example fix

# before
tools = [{'name': 'get_weather', 'parameters': {...}}]

# after
tools = [{'type': 'function',
          'function': {'name': 'get_weather', 'description': '...', 'parameters': {...}}}]

def check_tools(tools):
    for t in tools:
        assert isinstance(t, dict) and t.get('type') == 'function' and 'function' in t, f'bad tool schema: {t}'

def is_openai_tool_schema(t) -> bool:
    return isinstance(t, dict) and t.get('type') == 'function' and isinstance(t.get('function'), dict) and 'name' in t['function']

When it happens

Trigger: Passing tools=[{'name': 'foo', 'parameters': {...}}] (no type/function wrapper), or mixing in JSON-schema objects directly instead of the {'type':'function','function':{...}} envelope.

Common situations: Hand-building tool schemas; converting from other frameworks (Anthropic-style input_schema) without wrapping; passing toolkit.get_schemas() output of the wrong shape.

Related errors

• Invalid tool name '{mode}' in tool_choice.mode. Available to
• Input validation failed for tool '{tool_call.name}': {e.mess
• The injection template must contain the '{runtime_state}' pl
• Expected a 5-field cron expression, got {record.data.cron_ex
• Unsupported audio source type: {type(source)}

AI-assisted analysis of agentscope-ai/agentscope@e90f1c7592 (2026-08-28). Data as JSON: /api/errors/c1327fb1d39adefa. Report an issue: GitHub.