Inception Structured Output

structured_output.py

"""
Inception Structured Output
===========================

Cookbook example for `inception/structured_output.py`.
"""

from typing import List

from agno.agent import Agent, RunOutput  # noqa
from agno.models.inception import Inception
from pydantic import BaseModel, Field
from rich.pretty import pprint  # noqa

# ---------------------------------------------------------------------------
# Create Agent
# ---------------------------------------------------------------------------


class MovieScript(BaseModel):
    setting: str = Field(
        ..., description="Provide a nice setting for a blockbuster movie."
    )
    ending: str = Field(
        ...,
        description="Ending of the movie. If not available, provide a happy ending.",
    )
    genre: str = Field(
        ...,
        description="Genre of the movie. If not available, select action, thriller or romantic comedy.",
    )
    name: str = Field(..., description="Give a name to this movie")
    characters: List[str] = Field(..., description="Name of characters for this movie.")
    storyline: str = Field(
        ..., description="3 sentence storyline for the movie. Make it exciting!"
    )


# Agent that uses JSON mode (recommended for Inception).
# Inception's OpenAI-compatible endpoint does not advertise native
# `json_schema` structured outputs, so use_json_mode=True is the reliable path.
json_mode_agent = Agent(
    model=Inception(id="mercury-2"),
    description="You write movie scripts.",
    output_schema=MovieScript,
    use_json_mode=True,
)

# Agent that uses native structured outputs (output_schema without JSON mode).
structured_output_agent = Agent(
    model=Inception(id="mercury-2"),
    description="You write movie scripts.",
    output_schema=MovieScript,
)

# Get the response in a variable
# response: RunOutput = json_mode_agent.run("New York")
# pprint(response.content)

# ---------------------------------------------------------------------------
# Run Agent
# ---------------------------------------------------------------------------

if __name__ == "__main__":
    json_mode_agent.print_response("New York")

    structured_output_agent.print_response("New York")

The current Inception adapter disables native schema output by default, so both unmodified source agents request JSON Object mode. Mercury 2 now documents JSON Schema support. Apply the explicit opt-in below to make the second agent send a schema request; the source comments predate this provider capability.

output_schema describes the expected type. If parsing or validation fails, result.content can remain a string. Before accessing schema fields in a run result, use isinstance(result.content, YourSchema), replacing YourSchema with the class you passed as output_schema.

Run the Example

Set up your virtual environment

uv venv --python 3.12
source .venv/bin/activate

Install dependencies

uv pip install -U agno openai

Export your Inception API key

export INCEPTION_API_KEY="your_inception_api_key_here"

Enable the native schema comparison

Save the source as structured_output.py. In structured_output_agent, replace its model expression with Inception(id="mercury-2", supports_native_structured_outputs=True). Keep the first json_mode_agent and its use_json_mode=True setting unchanged.

Run the example

Save the code above as structured_output.py, then run:

python structured_output.py

Full source: cookbook/90_models/inception/structured_output.py