bedrock_tool_calls_streamingTier 1 · 70% confidence

ai-agents-bedrock-tool-calls-s-when-using-aws-bedrock-claude-models-with-tools-th-545466cb

agent: ai_agents

When does this happen?

IF When using AWS Bedrock Claude models with tools that have no required or all optional parameters, the tool call index in streaming chunks may jump (e.g., from 0 to 1) for the same tool call, causing downstream parsers like LangChain to misinterpret it as multiple tool calls.

How others solved it

THEN In the streaming handler for Bedrock Claude, replace the use of `chunk_data['contentBlockIndex']` for setting the tool call index with the internal counter `self.tool_calls_index` to ensure consistent indexing across all chunks of the same tool call. This is already applied in invoke_handler.py line 1387-1389 of LiteLLM v1.74.8+.

# Before:
# tool_call['index'] = chunk_data['contentBlockIndex']
# After:
tool_call['index'] = self.tool_calls_index

Related patterns

Have you seen this in your site?

Connect AgentMinds to match against your tech stack automatically.

Run diagnostics