Agent Using Multimodal Tool Response in Runs

Legacy DalleTools example that references a generated image in a follow-up run.

This legacy example keeps DALL-E image output in conversation history for follow-up runs.

DALL-E models are deprecated. This DalleTools example is retained as a legacy reference and no longer runs against the current OpenAI API. Use OpenAITools with GPT Image 2 in Image Generation Agent.

Code

agent_using_multimodal_tool_response_in_runs.py
from agno.agent import Agent
from agno.db.sqlite import SqliteDb
from agno.tools.dalle import DalleTools

# Create an Agent with the DALL-E tool
agent = Agent(
    tools=[DalleTools()],
    name="DALL-E Image Generator",
    add_history_to_context=True,
    db=SqliteDb(db_file="tmp/test.db"),
)

agent.print_response(
    "Generate an image of a Siamese white furry cat sitting on a couch?",
    markdown=True,
)

agent.print_response(
    "Which type of animal and the breed are we talking about?", markdown=True
)

Current Alternative

The source above uses removed DALL-E models and is preserved for reference. Follow Image Generation Agent for the current OpenAITools and GPT Image 2 pattern before rebuilding the multi-turn flow.