Fal | Agno

FalTools enable an Agent to perform media generation tasks.

Prerequisites

The following example requires the fal_client library and an API key which can be obtained from Fal.

uv pip install -U agno fal_client openai

export FAL_KEY=***

FalTools checks the FAL_API_KEY environment variable and logs an error when it is missing. Authentication still works with FAL_KEY alone because fal_client reads it directly. Set both variables to avoid the log message.

Example

The following agent will use FAL to generate any video requested by the user.

cookbook/91_tools/fal_tools.py

from agno.agent import Agent
from agno.models.openai import OpenAIResponses
from agno.tools.fal import FalTools

fal_agent = Agent(
    name="Fal Video Generator Agent",
    model=OpenAIResponses(id="gpt-5.2"),
    tools=[FalTools(model="fal-ai/hunyuan-video")],
    description="You are an AI agent that can generate videos using the Fal API.",
    instructions=[
        "When the user asks you to create a video, use the `generate_media` tool to create the video.",
        "Return the URL as raw to the user.",
        "Don't convert video URL to markdown or anything else.",
    ],
    markdown=True,
    debug_mode=True,
)

fal_agent.print_response("Generate video of balloon in the ocean")

Toolkit Params

Parameter	Type	Default	Description
`api_key`	`str`	`None`	Stored by Agno; not forwarded to the module-level Fal client. Set `FAL_KEY` for SDK authentication.
`model`	`str`	`fal-ai/hunyuan-video`	The model to use for the media generation.
`enable_generate_media`	`bool`	`True`	Enable the generate_media functionality.
`enable_image_to_image`	`bool`	`False`	Enable the image_to_image functionality.
`all`	`bool`	`False`	Enable all functionality.

Toolkit Functions

Function	Description
`generate_media`	Generate media using the configured model; the model determines the output type.
`image_to_image`	Transform an input image using the fixed Flux image-to-image endpoint, independently of the constructor model.

Developer Resources

Tools