BerriAI/litellm · error · OCIError

Tool call `function` must be a dictionary

Error message

Tool call `function` must be a dictionary

What it means

Inside each function tool call, the `function` field must be a dict containing name and arguments. If it is missing, None, a string (double-encoded JSON), or any non-dict, OCI translation aborts with OCIError 400. Mirrors the OpenAI tool-call structure that OCI GENERIC maps onto OCIToolCall.

Solutions

1. Keep `function` a dict: {"name": ..., "arguments": "<json str>"}
2. Repair double-encoded history at load: json.loads(tc["function"]) when it is a str
3. Echo the structure exactly as returned by prior completion responses
4. Validate history shape (see type guard) before the first OCI call

Example fix

# before
{"id":"1","type":"function","function":"{\"name\":\"get_weather\",\"arguments\":\"{}\"}"}

# after
{"id":"1","type":"function","function":{"name":"get_weather","arguments":"{}"}}

import json

def normalize_tool_call_function(tool_calls: list) -> list:
    out = []
    for tc in tool_calls:
        fn = tc.get("function")
        if isinstance(fn, str):
            tc = {**tc, "function": json.loads(fn)}
        out.append(tc)
    return out

def tool_call_function_is_dict(tc) -> bool:
    return isinstance(tc.get("function"), dict)

When it happens

Trigger: Assistant history with tool_calls=[{"id":"1","type":"function","function":"{\"name\":\"f\"}"}] (function JSON-stringified) or with the function key omitted, sent to an oci/generic.* model.

Common situations: json.dumps applied one level too deep when persisting history; hand-crafted tool_calls from an LLM's raw text output parsed shallowly; frameworks storing function as a flat string for compactness; partial dict after aggressive key filtering.

Related errors

• Each tool call must be a dictionary
• Message `tool_calls` must be a list
• OCI only supports function tool calls
• Tool call `function.arguments` must be a JSON string
• Tool call `function.name` must be a string

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/34bbc070b088fe9c. Report an issue: GitHub.

Appendix: source

Thrown at litellm/llms/oci/chat/generic.py:129

    )


def adapt_messages_to_generic_oci_standard_tool_call(role: str, tool_calls: list) -> OCIMessage:
    """Convert an assistant tool-call message to OCI format."""
    tool_calls_formatted: Final = []
    for tool_call in tool_calls:
        if not isinstance(tool_call, dict):
            raise OCIError(status_code=400, message="Each tool call must be a dictionary")
        if tool_call.get("type") != "function":
            raise OCIError(status_code=400, message="OCI only supports function tool calls")

        tool_call_id = tool_call.get("id")
        if not isinstance(tool_call_id, str):
            raise OCIError(status_code=400, message="Tool call `id` must be a string")

        tool_function = tool_call.get("function")
        if not isinstance(tool_function, dict):
            raise OCIError(status_code=400, message="Tool call `function` must be a dictionary")

        function_name = tool_function.get("name")
        if not isinstance(function_name, str):
            raise OCIError(status_code=400, message="Tool call `function.name` must be a string")

        arguments = tool_call["function"].get("arguments", "{}")
        if not isinstance(arguments, str):
            raise OCIError(
                status_code=400,
                message="Tool call `function.arguments` must be a JSON string",
            )

        tool_calls_formatted.append(
            OCIToolCall(
                id=tool_call_id,
                type="FUNCTION",
                name=function_name,
                arguments=arguments,

View on GitHub (pinned to 77b7c6c40c)