[FEATURE] Support adding prompts to conversational history when using structured output

[dwu12]

Problem Statement

I would like Strands to add prompt to conversational history when using structured output method. I followed the example code:

agent = Agent()

# Build up conversation context
agent("What do you know about Paris, France?")
agent("Tell me about the weather there in spring.")

# Extract structured information with a prompt
class CityInfo(BaseModel):
    city: str
    country: str
    population: Optional[int] = None
    climate: str

# Uses existing conversation context with a prompt
result = agent.structured_output(CityInfo, "Extract structured information about Paris")

However the agent messages only contain the first two queries: "What do you know about Paris, France?", "Tell me about the weather there in spring." without adding the prompt "Extract structured information about Paris" in the history.

Can we add this feature to the history? For example convert structured class into json file and store both (prompt and json output to the conversational history)

Proposed Solution

No response

Use Case

See above. For example, for use case requires query decomposition + query rewriting as the first step, I would like directly use agent.structured_output() to generate sub queries. However, If I have a follow up question, it won't work because the previous prompt wasn't stored in conversational history.

Alternatives Solutions

No response

Additional Context

No response

[aditya270520]

Resolves #810

Problem Statement

Agent structured_output() method did not add prompts to conversational history, making it impossible for follow-up questions to reference previous structured output.

Solution

• Modified structured_output_async to add prompts to conversation history when provided
• Modified structured_output_async to add structured output results to conversation history
• Updated all related tests to verify the new behavior
• Maintains backward compatibility

Testing

• ✅ All 147 agent tests pass
• ✅ Comprehensive end-to-end testing completed
• ✅ Edge cases handled
• ✅ No linting issues

Example Usage

agent = Agent()
agent("What do you know about Paris, France?")
agent("Tell me about the weather there in spring.")

# Now works - prompt and structured output are added to conversation history
result = agent.structured_output(CityInfo, "Extract structured information about Paris")

# Follow-up questions can now reference the structured output!

The implementation is production-ready and fully tested! 🚀

[xylophone21]

When will this patch merge into master?