[Fix] Fix for null tool call error

[Sidharthenee333]

🐞 Bug Fix

📋 Summary

This PR fixes a bug in the inject_history_multiturn function in the llm_node that caused crashes when processing assistant messages without tool calls in multi-turn conversations.

Issue:

When an assistant response in chat history didn't contain tool calls (empty list []), the code attempted to convert the empty list to LangChain tool calls, causing a ValueError: Expected 'function' key in OpenAI tool call error in the
convert_openai_to_langchain_toolcall function.

Fix:

Added validation checks to ensure tool_calls_data contains items before attempting conversion:

🔧 Impacted Components

• 🧠 llm_node.inject_history_multiturn — method now correctly handles assistant messages without tool calls

🧪 How to Test

1. Run an end-to-end web/desktop agent eval task using any claude model with chat history injection enabled
2. Note than in the output json file for any record even though there is a null tool call during any retry the subsequent retries have valid tool call and do not have an server errors.

📸 Screenshots

N/A

✅ Checklist

• Lint fixes and unit testing done
• End to end task testing
• Documentation updated

Notes

N/A

[psriramsnc]

LGTM 🚀