Fix mypy regression: TypedDict key error in fireworks_ai transformation

[ghost]

Regression Fix

Failing Job: mypy_linting
Caused By: #20334 (commit 3765d88)
Author: @sameer-berri

What Broke

PR #20334 added code to remove provider_specific_fields from messages before sending to FireworksAI. The code calls message.pop("provider_specific_fields", None) where message is typed as AllMessageValues (a Union of TypedDicts).

mypy correctly complained that none of the TypedDicts in the union have a provider_specific_fields key:

llms/fireworks_ai/chat/transformation.py:242: error: TypedDict "ChatCompletionUserMessage" has no key "provider_specific_fields"  [typeddict-item]
llms/fireworks_ai/chat/transformation.py:242: error: TypedDict "ChatCompletionAssistantMessage" has no key "provider_specific_fields"  [typeddict-item]
... (6 total errors)

This Fix

Cast message to dict before calling pop(). This is safe because:

1. The code already checks isinstance(message, dict) before reaching this line
2. The code already checks "provider_specific_fields" in message before popping

This is consistent with line 239 which already uses cast(dict, message) for filter_value_from_dict().

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
litellm	Ready	Preview, Comment	Feb 4, 2026 4:46am

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.

---

Shin seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[greptile-apps]

Greptile Overview

Greptile Summary

This PR fixes a mypy type error introduced in PR #20334. The error occurred because message.pop("provider_specific_fields", None) was called on a variable typed as AllMessageValues (a Union of TypedDicts), none of which define the provider_specific_fields key.

The fix adds cast(dict, message) before calling .pop(), which is safe because:

• The code already validates isinstance(message, dict)
• The code checks "provider_specific_fields" in message before popping
• This pattern is consistent with line 238 which uses cast(dict, message) for filter_value_from_dict()

This is a minimal, type-safe fix that resolves the mypy regression without changing runtime behavior.

Confidence Score: 5/5

• This PR is safe to merge - it's a minimal type annotation fix with no runtime changes
• The change only adds a type cast to satisfy mypy. The runtime behavior is unchanged, proper validation is already in place, and the pattern matches existing code on line 238
• No files require special attention

Important Files Changed

Filename	Overview
litellm/llms/fireworks_ai/chat/transformation.py	Added cast(dict, message) to fix mypy type error when calling .pop() on AllMessageValues union type

Sequence Diagram

sequenceDiagram
    participant Caller
    participant FireworksAIConfig
    participant TypeChecker as mypy Type Checker
    
    Caller->>FireworksAIConfig: _transform_messages_helper(messages)
    Note over FireworksAIConfig: messages: List[AllMessageValues]
    
    loop For each message
        FireworksAIConfig->>FireworksAIConfig: Check if isinstance(message, dict)
        FireworksAIConfig->>FireworksAIConfig: Check if "provider_specific_fields" in message
        
        alt Has provider_specific_fields
            Note over FireworksAIConfig,TypeChecker: Before fix: message.pop() fails type check
            Note over FireworksAIConfig,TypeChecker: After fix: cast(dict, message).pop() passes
            FireworksAIConfig->>FireworksAIConfig: cast(dict, message).pop("provider_specific_fields", None)
        end
    end
    
    FireworksAIConfig-->>Caller: Return cleaned messages

Loading

[greptile-apps]

_{1 file reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}