BerriAI/litellm · error · ValueError

Advisor tool must have a valid model

Error message

Advisor tool must have a valid model

What it means

For the Anthropic advisor tool (ANTHROPIC_ADVISOR_TOOL_TYPE), LiteLLM requires tool['model'] to be a string naming the advisor model; it builds AnthropicAdvisorTool(name='advisor', model=...). A missing or non-string 'model' raises this ValueError client-side.

Source

Thrown at litellm/llms/anthropic/chat/transformation.py:759

        elif tool["type"] == "tool_search_tool_bm25_20251119":
            # Tool search tool using BM25
            from litellm.types.llms.anthropic import AnthropicToolSearchToolBM25

            tool_name_obj = tool.get("name", "tool_search_tool_bm25")
            if not isinstance(tool_name_obj, str):
                raise ValueError("Tool search tool must have a valid name")
            tool_name = tool_name_obj
            returned_tool = AnthropicToolSearchToolBM25(
                type="tool_search_tool_bm25_20251119",
                name=tool_name,
            )
        elif tool["type"] == ANTHROPIC_ADVISOR_TOOL_TYPE:
            from litellm.types.llms.anthropic import AnthropicAdvisorTool

            _tool_dict: Final = cast(dict, tool)
            advisor_model: Final = _tool_dict.get("model")
            if not isinstance(advisor_model, str):
                raise ValueError("Advisor tool must have a valid model")
            _advisor_tool: Final = AnthropicAdvisorTool(
                type=ANTHROPIC_ADVISOR_TOOL_TYPE,
                name="advisor",
                model=advisor_model,
            )
            if _tool_dict.get("max_uses") is not None:
                _advisor_tool["max_uses"] = _tool_dict["max_uses"]
            if _tool_dict.get("caching") is not None:
                _advisor_tool["caching"] = _tool_dict["caching"]
            returned_tool = _advisor_tool
        if returned_tool is None and mcp_server is None:
            raise ValueError(f"Unsupported tool type: {tool['type']}")

        ## check if cache_control is set in the tool
        _cache_control: Final = tool.get("cache_control", None)
        _cache_control_function: Final = tool.get("function", {}).get("cache_control", None)
        if returned_tool is not None:
            # Only set cache_control on tools that support it (not tool search tools)

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Add 'model': '<model-id>' (a string, e.g. the advisor model Anthropic documents) to the advisor tool dict.
2. Verify optional extras (max_uses, caching) are separate keys; only 'model' is mandatory.
3. Default empty strings from config to a concrete model id before building the tool.

Example fix

# before
{"type": "advisor_20251119", "max_uses": 2}

# after
{"type": "advisor_20251119", "model": "claude-sonnet-4-5", "max_uses": 2}

if tool.get("type") == "advisor_20251119":
    m = tool.get("model")
    if not isinstance(m, str) or not m:
        raise ValueError("advisor tool requires a non-empty string 'model'")

def advisor_tool_valid(tool) -> bool:
    return isinstance(tool.get("model"), str) and len(tool["model"]) > 0

When it happens

Trigger: Passing {'type': <advisor type>, 'max_uses': 2} without 'model', or with 'model': None / a dict. The advisor tool always needs the backing model specified.

Common situations: Enabling the advisor feature from release notes without copying the full required fields; environment-driven configs where the model variable was empty and serialized as None.

Related errors

• advisor tool definition must include a 'model' field specify
• Missing required parameter: parameters
• Missing required parameter: display_width_px or display_heig
• Missing required parameter: name
• Tool search tool must have a valid name

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/8127354fae33233a. Report an issue: GitHub.