LiteLLM tool call ids carry the embedded Gemini thought signature

[benclarkeio]

🔴 Required Information

Describe the Bug:

LiteLLM embeds a Gemini thought_signature inside the tool call id, separated by __thought__ (see _THOUGHT_SIGNATURE_SEPARATOR). _message_to_generate_content_response already extracts that signature onto part.thought_signature, but then assigns the raw id to part.function_call.id:

thought_signature = _extract_thought_signature_from_tool_call(tool_call)
part = types.Part.from_function_call(...)
part.function_call.id = tool_call.id      # still "call_abc__thought__AY89a18..."
if thought_signature:
    part.thought_signature = thought_signature

So every consumer of function_call_id gets several hundred characters of base64 glued onto the real id:

call_abc__thought__AY89a18qpllj6mkpxPjhFEcbBbtEsqy4Ia4Eam2lD_NsoZ...

In practice it leaks into logs, into any UI that displays a tool call id, and into anything that derives a name from one.

Steps to Reproduce:

1. Run an agent on LiteLlm(model="vertex_ai/gemini-3.5-flash") (any Gemini model through LiteLLM) with a tool.
2. Prompt it so the model both thinks and calls the tool.
3. Inspect part.function_call.id on the resulting event, or anything keyed by it.

Expected Behavior:

part.function_call.id is the id LiteLLM assigned, e.g. call_abc, with the signature available separately on part.thought_signature.

Observed Behavior:

part.function_call.id is call_abc__thought__AY89a18..., carrying a few hundred characters of base64 that every downstream consumer inherits.

Environment Details:

• ADK Library Version (pip show google-adk): 1.31.1, and current main
• Desktop OS: macOS
• Python Version (python -V): 3.13.5

Model Information:

• Are you using LiteLLM: Yes
• Which model is being used: vertex_ai/gemini-3.5-flash