Output Formatting in Tool Responses: Is My Approach Correct?

[mrctito]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

@tool
async def tool_fetch_promotions_or_offers(customer_code: str):
    """
    Useful for fetching promotions or offers available to the customer.
    """

    output_format = dedent("""
    <START INSTRUCTION>When the number of promotional products is greater than 0, print the message below:<END INSTRUCTION>
                        
    "Products on promotion today:

    1. product_name (in bold)
    - Code: product_code_sku
    - Brand: brand
    - Packaging: packaging_code
    - Unit price: price
    - Discount: discount_percentage

    [Skip a line and continue the list until ALL promotional products are printed!!!]
                        
    Take advantage of the promotions!
    """)

    result = await fetch_promotions_or_offers(customer_code)
    result = {"status": "SUCCESS", "Instructions": output_format, "List of products with offers or promotions": result}
    return result

Description

Hello @dosu,

I have my multi-agent project that uses various tools, using create_react_agent, and to format the output that the LLM will present, I include a formatting instruction in the response. Is the correct way to do it???

Here is an example of how I format the output in the tool:

@tool
async def tool_fetch_promotions_or_offers(customer_code: str):
"""
Useful for fetching promotions or offers available to the customer.
"""

# here is the formatting instructions 
output_format = dedent("""
<START INSTRUCTION>When the number of promotional products is greater than 0, print the message below:<END INSTRUCTION>
                    
"Products on promotion today:

1. product_name (in bold)
- Code: product_code_sku
- Brand: brand
- Packaging: packaging_code
- Unit price: price
- Discount: discount_percentage

[Skip a line and continue the list until ALL promotional products are printed!!!]
                    
Take advantage of the promotions!
""")

result = await fetch_promotions_or_offers(customer_code)
result = {"status": "SUCCESS", "Instructions": output_format, "List of products with offers or promotions": result}
return result

Thank you

System Info

langchain==0.2.6
langchain-community==0.2.6
langchain-core==0.2.10
langchain-experimental==0.0.62
langchain-openai==0.1.13
langchain-qdrant==0.1.0
langchain-text-splitters==0.2.1

[dosubot[bot]]

Yes, including a formatting instruction in the response of a tool using create_react_agent is the correct way to format the output that the LLM will present. The LangChain documentation specifies the expected output formats for actions and final answers, which are used to parse the LLM's output correctly ^[1][2][3].

Additionally, the LangChain documentation provides examples of formatting instructions within tool responses for multi-agent projects. The FORMAT_INSTRUCTIONS section demonstrates how to format responses when interacting with tools or responding directly to the user ^[4][5][6].

Your approach of including detailed formatting instructions in the tool's response is aligned with the recommended practices.

To continue talking to Dosu, mention @dosu.