Agents with different tools not working on bedrock

tomaszbk commented 1 month ago

Checked other resources

[X] I added a very descriptive title to this issue.
[X] I searched the LangGraph/LangChain documentation with the integrated search.
[X] I used the GitHub search to find a similar question and didn't find it.
[X] I am sure that this is a bug in LangGraph/LangChain rather than my code.
[X] I am sure this is better as an issue rather than a GitHub discussion, since this is a LangGraph bug and not a design question.

Example Code

from langchain_aws import ChatBedrockConverse

llm = ChatBedrockConverse(
        model=AWS_BEDROCK_MODEL,
        region_name=AWS_BEDROCK_REGION,
    )

orchestration_tools = [choose_action]
orchestration_tool_node = ToolNode(orchestration_tools)
orchestration_model = llm.bind_tools(orchestration_tools)

conversation_model = llm

products_tools = [get_products, get_product_details]
products_tools_node = ToolNode(products_tools)
products_model = llm.bind_tools(products_tools)

purchase_tools = [make_purchase, get_user_data]
purchase_tools_node = ToolNode(purchase_tools)
purchase_model = llm.bind_tools(purchase_tools)

Error Message and Stack Trace (if applicable)

botocore.errorfactory.ValidationException: An error occurred (ValidationException) when calling the Converse operation: The toolConfig field must be defined when using toolUse and toolResult content blocks.

Description

Some agents have tools binded and others doesn't. When the chat history with tool calls is received in an agent without tools (or don't know if same would happen in one that has tools but different ones), the specified error appears

System Info

"boto3>=1.35.29",
"fastapi>=0.115.0",
"langchain-aws>=0.2.1",
"langgraph>=0.2.28",
"pydantic-settings>=2.5.2",
"pydantic>=2.9.2",
"python-multipart>=0.0.11",
"uvicorn>=0.31.0"

j10sanders commented 1 month ago

Same issue for me, without using langchain - just using bedrock sdk. So it's a bedrock issue unfortunately

dlqqq commented 1 month ago

Can somebody please transfer this issue to langchain-ai/langchain-aws?

langchain-ai / langchain